Sorption of Cr(III) in fixed-bed column by modified pine ...

93
Kevin Neves Sorption of Cr(III) in fixed-bed column by modified pine bark: optimization and modelling studies Master‘s thesis in the scientific area of Chemical Engineering, supervised by Professor Doctor Licínio Ferreira and Professor Doctor Margarida Quina and submitted to the Department of Chemical Engineering, Faculty of Science and Technology, University of Coimbra September 2017

Transcript of Sorption of Cr(III) in fixed-bed column by modified pine ...

Page 1: Sorption of Cr(III) in fixed-bed column by modified pine ...

Kevin Neves

Sorption of Cr(III) in fixed-bed column by modified pine bark

optimization and modelling studies

Masterlsquos thesis in the scientific area of Chemical Engineering supervised by Professor Doctor Liciacutenio Ferreira and

Professor Doctor Margarida Quina and submitted to the Department of Chemical Engineering Faculty of Science and Technology

University of Coimbra

September 2017

i

Kevin Neves

Sorption of Cr(III) in fixed-bed column by

modified pine bark optimization and modelling

studies

Master Thesis in the scientific area of Chemical Engineering and submitted to the Department of

Chemical Engineering Faculty of Science and Technology University of Coimbra

Supervisor(es)Supervisor(s)

Prof Dr Liciacutenio Manuel Gando de Azevedo Ferreira

Prof Dr Margarida Maria Joatildeo de Quina

Host institutions

CIEPQPF - Research Centre for Chemical Processes Engineering and Forest Products

Department of Chemical Engineering Faculty of Sciences and Technology of the University of

Coimbra

Coimbra September 2017

i

ACKNOWLEDGEMENTS

The time has come one more step completed It is time to embrace the future enter in a

new world of discoveries and challenges Its time to look into the past with a smile on my face

remembering all the moments Ive pass through

First of all I want to thank my parents and my grandmother for all the support help

and encouragement along this path without them it would not be possible To my uncles to my

godmother Joana and to my grandparents also a sincere gratitude

To my supervisors Professor Doctor Liciacutenio Ferreira and Professor Doctor Margarida

Quina my deepest acknowledgment for all the support availability and guidance in all the

moments during this project

To Aline for her kindness availability cooperation guidance and advice during this

semester at the laboratory and also outside of it

To my university family Rita Xana Joana Priscila Vera Miguel Rodrigo Dani and

Joli who have been always there for all the moments and for their friendship Without them it

would be much harder

To all my friends of Mala who always believed in me for their friendship and all the

moments that wersquove passed

To all the Professors of the Chemical Engineering Department of the University of

Coimbra who provided me all the knowledge necessary to complete this course

To all of you my deepest and sincere acknowledgment

iii

ABSTRACT

The adsorption process in fixed bed columns is one of the most advantageous processes

for the removal of heavy metals allowing the treatment of large volumes of effluent The fact

that the bed can be regenerated and the possibility of being reused allowing the recover the

adsorbed metal are other factors in favor of its use in large-scale industrial processes Chromium

is the heavy metal selected in this study because of its relevance in the electroplating coating

and tanning industries The effluents from these industries may contain detrimental

concentrations of Cr

This work aims to analyse the dynamic behavior of a fixed bed column in which the

adsorbent was prepared in the lab Following the previous work mercerized pine bark (MPB)

will be used as adsorbent to remove Cr (III) and to purify a synthetic effluent Thus operational

parameters such as bed height fluid velocity and effluent inlet concentration were tested to

obtain the maximum saturated bed fraction and to predict the behavior of the adsorption process

for each case A mechanistic model was developed for predicting the behaviour of the fixed

bed that includes equilibrium parameters and the Linear Driving Force (LDF) approach to

express the intraparticular mass transport For an inlet concentration of 100 mgL the maximum

saturated bed fraction was 065 obtained for the case where a height of 65 cm and a superficial

velocity of 1443times10-2 cms is used Moreover the influence of the same parameters was also

studied to minimize the difference between the exhaust time and the break time of the column

In this case using an inlet concentration of 100 mgL a bed height of 25 cm and a surface

speed of 1443times10-2 cms were the best conditions

Moreover a Box-Benken-Design was used to determine the most influent inlet variables

(factors) on adsorption efficiency By using multiple linear regression empirical models were

developed and the optimal operating conditions to maximize the saturated bed fraction were

identified Finally a regeneration of the adsorbent material was assessed through two chemical

agents (nitric acid and NN-Bis(carboxymethyl)-DL-alanine trisodium salt)

From this study it was possible to conclude that the biosorbent tested can be used in

fixed-bed column process for remove chromium from liquid effluents

Keywords Adsorption chromium fixed-bed column modified pine bark mechanistic model

Box-Benken-Design regeneration

v

RESUMO

O processo de adsorccedilatildeo utilizando colunas de leito fixo mostra-se como um dos

processos mais vantajosos para a remoccedilatildeo de metais pesados de efluentes industriais O facto

de o leito poder ser regenerado haver a possibilidade de ser reutilizado e recuperar o metal

adsorvido satildeo outros fatores a favor para o seu uso O croacutemio por ser utilizado em processos

de galvanoplastia revestimentos e curtumes ocorre frequentemente no meio ambiente devido

aos efluentes provenientes deste tipo de induacutestrias

Neste trabalho seraacute estudado o comportamento dinacircmico de uma coluna de leito fixo

em que o adsorvente utilizado foi obtido no laboratoacuterio a partir de casca de pinheiro No

seguimento de trabalhos anteriores este material seraacute utilizado como adsorvente para remover

o Cr(III) Foram realizadas vaacuterias experiecircncias para testar diversos paracircmetros operacionais

como a altura do leito a velocidade superficial do fluido e a concentraccedilatildeo de entrada do

efluente Na praacutetica pretende-se obter a maacutexima fraccedilatildeo de leito saturado e prever o

comportamento dinacircmico do processo de adsorccedilatildeo Um modelo de natureza mecaniacutestica foi

desenvolvido de modo a prever a resposta do sistema face a variaccedilotildees dos paracircmetros

operacionais Foram incluiacutedas as relaccedilotildees de equiliacutebrio bem como o modelo Linear Driving

Force (LDF) para expressar o transporte de massa intraparticular Deste modo verificou-se que

para uma concentraccedilatildeo de entrada fixa em 100 mgL a maacutexima fraccedilatildeo de leito saturado obtida

eacute de 065 para o caso onde se utilizada uma altura de leiro de 65 cm e uma velocidade

superficial de 1443times10-2 cms Foi tambeacutem estudado a influecircncia dos mesmos paracircmetros de

modo a minimizar a diferenccedila entre o tempo de exaustatildeo onde as condiccedilotildees oacutetimas satildeo uma

altura de leito de 25 cm e uma velocidade superficial de 1443times10-2 cms

Para aleacutem disto desenho fatorial do tipo Box-Benken-Design foi utilizado de modo a

determinar quais as variaacuteveis que mais influenciam a eficiecircncia do processo de adsorccedilatildeo em

leito fixo Utilizando uma regressatildeo linear muacuteltipla dois modelos empiacutericos foram construiacutedos

onde foram identificas as condiccedilotildees oacutetimas de operaccedilatildeo de modo a maximizar a fraccedilatildeo de leito

saturado Por fim a regeneraccedilatildeo do adsorvente foi feita utilizando dois agentes quiacutemicos (aacutecido

niacutetrico e NN-Bis(carboxymethyl)-DL-alanine trisodium salt)

Assim deste estudo foi possiacutevel concluir que o adsorvente testado pode ser utilizado em

processos com colunas de leito fixo para a remoccedilatildeo de croacutemio de efluentes liacutequidos

Palavras chave Adsorccedilatildeo croacutemio coluna de leito fixo casca de pinheiro modificada modelo

mecaniacutestico Box-Benken Design regeneraccedilatildeo

vii

INDEX

Acknowledgements i

Abstract iii

Resumo v

List of Figures ix

List of Tables xi

Acronyms xv

1 Introduction 1

Work motivation and scope 1

Thesis objective 1

Thesis structure 2

2 Theoretical background 3

21 Heavy metals as contaminants for the environment 3

22 Methods for removing heavy metals from wastewaters 5

23 Adsorption process 7

24 Factors affecting adsorption processes 8

25 Desorption and regeneration of adsorbents 9

26 Adsorbents 11

261 Properties of the adsorbent 11

262 Low-cost adsorbents 11

27 Fixed bed adsorption 15

271 The breakthrough concept 16

272 Mathematical models of fixed bed adsorption 19

3 State of the art 27

31 Heavy metal removal in fixed bed column 27

32 Mathematical models for fixed bed adsorption 29

4 Materials and methods 33

41 Materials 33

411 Adsorbate 33

412 Adsorbent 33

viii

42 Experimental methods 34

421 Column experiment 34

43 Analytical techniques 35

431 Dynamic Light Scattering (DLS) 35

432 Moisture of the adsorbent 35

433 Bed porosity analysis 35

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF) 36

44 Experimental planning 36

5 Results and discussion 37

51 Experimental and calculated breakthrough curves 37

511 Effect of fluid superficial velocity on breakthrough curves 37

512 Effect of bed depth on breakthrough curves 40

513 Effect of feed concentration on breakthrough curves 41

52 Comparison of the simulation model with the experimental results 43

53 Statistical treatment (DOE) 45

54 Regeneration test 50

55 Validation of the empirical models 53

6 Conclusion and future work 57

REFERENCES 59

7 Appendixes I

71 Appendix A- Particle size distribution I

72 Appendix B- Experimental data II

ix

LIST OF FIGURES

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12) 4

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6

M- (adapted from11) 5

Figure 23 Stages of the adsorption process- (adapted from20 ) 7

Figure 24 Concentration and temperature profiles in adsorption and desorption processes-

(adapted from18) 10

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29) 12

Figure 26 Molecular structure of lignin- (adapted from30) 13

Figure 27 Molecular structure of cellulose- (adapted from31) 13

Figure 28 Pine bark from (Pinus pinaster) 15

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of

operation ndash (adapted from42) 16

Figure 210 Typical representation of a breakthrough curve 17

Figure 211 Breakthrough curve for plug flow in the bed 19

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

21

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature =

25C and pH 5)- (adapted from44) 23

Figure 41 Fixed-bed column apparels used in experiments 34

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

superficial velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm 38

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

bed depths (a) 119906119900=962times10-2 cms (b) 119906119900=1443times10-2 cms and (c) 119906119900=1924times10-2 cms 40

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III)

adsorption in MPB using different feed concentrations 42

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25

cm (b) H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough

curves (982 202 and 302 mgL) 44

Figure 55 Pareto chart of the factors affecting FLS 45

Figure 56 Observed vs predicted values 46

Figure 57 Three-dimensional response surface effect of superficial velocity and height on

the fraction of saturated bed (FLS) 47

Figure 58 Pareto chart of the factors affecting Δtads 48

x

Figure 59 Observed vs predicted values 49

Figure 510 Three-dimensional response surface effect of superficial velocity and height on

Δtads 50

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by

the simulated model 51

Figure 512 Desorption curves 52

xi

LIST OF TABLES

Table 21 The maximum concentration limits for hazardous heavy metals in industrial

effluents- (adapted from8) 3

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted

from16) 6

Table 31 Column studies on adsorption of heavy metals 28

Table 32 General dynamic models for column adsorption process 30

Table 33 More detailed general dynamic models for column adsorption studies 30

Table 34 Additional models to express adsorbate-adsorbent interactions 31

Table 35 Empirical models for column adsorption studies 32

Table 41 Physical properties of MPB 33

Table 42 Parameters and factor levels 36

Table 51 Factorial design of the experiences performed 37

Table 52 Experimental conditions and calculated parameters to assess the process

performance 39

Table 53 Experimental conditions and calculated parameters to assess the process

performance 43

Table 54 Properties and parameter values used in the simulation 44

Table 55 Experimental results vs predicted results 47

Table 56 Experimental results vs predicted results 49

Table 57 Experimental results obtained from the saturation of the column 51

Table 58 Adsorbed and desorbed amount of chromium 52

Table 59 Experimental values vs predicted values for the model built for the case in which

the FLS was considered as dependent variable 54

Table 510 Experimental values vs predicted values for the model built for the case in which

the (tE-tbp) was considered as dependent variable 54

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 2: Sorption of Cr(III) in fixed-bed column by modified pine ...

i

Kevin Neves

Sorption of Cr(III) in fixed-bed column by

modified pine bark optimization and modelling

studies

Master Thesis in the scientific area of Chemical Engineering and submitted to the Department of

Chemical Engineering Faculty of Science and Technology University of Coimbra

Supervisor(es)Supervisor(s)

Prof Dr Liciacutenio Manuel Gando de Azevedo Ferreira

Prof Dr Margarida Maria Joatildeo de Quina

Host institutions

CIEPQPF - Research Centre for Chemical Processes Engineering and Forest Products

Department of Chemical Engineering Faculty of Sciences and Technology of the University of

Coimbra

Coimbra September 2017

i

ACKNOWLEDGEMENTS

The time has come one more step completed It is time to embrace the future enter in a

new world of discoveries and challenges Its time to look into the past with a smile on my face

remembering all the moments Ive pass through

First of all I want to thank my parents and my grandmother for all the support help

and encouragement along this path without them it would not be possible To my uncles to my

godmother Joana and to my grandparents also a sincere gratitude

To my supervisors Professor Doctor Liciacutenio Ferreira and Professor Doctor Margarida

Quina my deepest acknowledgment for all the support availability and guidance in all the

moments during this project

To Aline for her kindness availability cooperation guidance and advice during this

semester at the laboratory and also outside of it

To my university family Rita Xana Joana Priscila Vera Miguel Rodrigo Dani and

Joli who have been always there for all the moments and for their friendship Without them it

would be much harder

To all my friends of Mala who always believed in me for their friendship and all the

moments that wersquove passed

To all the Professors of the Chemical Engineering Department of the University of

Coimbra who provided me all the knowledge necessary to complete this course

To all of you my deepest and sincere acknowledgment

iii

ABSTRACT

The adsorption process in fixed bed columns is one of the most advantageous processes

for the removal of heavy metals allowing the treatment of large volumes of effluent The fact

that the bed can be regenerated and the possibility of being reused allowing the recover the

adsorbed metal are other factors in favor of its use in large-scale industrial processes Chromium

is the heavy metal selected in this study because of its relevance in the electroplating coating

and tanning industries The effluents from these industries may contain detrimental

concentrations of Cr

This work aims to analyse the dynamic behavior of a fixed bed column in which the

adsorbent was prepared in the lab Following the previous work mercerized pine bark (MPB)

will be used as adsorbent to remove Cr (III) and to purify a synthetic effluent Thus operational

parameters such as bed height fluid velocity and effluent inlet concentration were tested to

obtain the maximum saturated bed fraction and to predict the behavior of the adsorption process

for each case A mechanistic model was developed for predicting the behaviour of the fixed

bed that includes equilibrium parameters and the Linear Driving Force (LDF) approach to

express the intraparticular mass transport For an inlet concentration of 100 mgL the maximum

saturated bed fraction was 065 obtained for the case where a height of 65 cm and a superficial

velocity of 1443times10-2 cms is used Moreover the influence of the same parameters was also

studied to minimize the difference between the exhaust time and the break time of the column

In this case using an inlet concentration of 100 mgL a bed height of 25 cm and a surface

speed of 1443times10-2 cms were the best conditions

Moreover a Box-Benken-Design was used to determine the most influent inlet variables

(factors) on adsorption efficiency By using multiple linear regression empirical models were

developed and the optimal operating conditions to maximize the saturated bed fraction were

identified Finally a regeneration of the adsorbent material was assessed through two chemical

agents (nitric acid and NN-Bis(carboxymethyl)-DL-alanine trisodium salt)

From this study it was possible to conclude that the biosorbent tested can be used in

fixed-bed column process for remove chromium from liquid effluents

Keywords Adsorption chromium fixed-bed column modified pine bark mechanistic model

Box-Benken-Design regeneration

v

RESUMO

O processo de adsorccedilatildeo utilizando colunas de leito fixo mostra-se como um dos

processos mais vantajosos para a remoccedilatildeo de metais pesados de efluentes industriais O facto

de o leito poder ser regenerado haver a possibilidade de ser reutilizado e recuperar o metal

adsorvido satildeo outros fatores a favor para o seu uso O croacutemio por ser utilizado em processos

de galvanoplastia revestimentos e curtumes ocorre frequentemente no meio ambiente devido

aos efluentes provenientes deste tipo de induacutestrias

Neste trabalho seraacute estudado o comportamento dinacircmico de uma coluna de leito fixo

em que o adsorvente utilizado foi obtido no laboratoacuterio a partir de casca de pinheiro No

seguimento de trabalhos anteriores este material seraacute utilizado como adsorvente para remover

o Cr(III) Foram realizadas vaacuterias experiecircncias para testar diversos paracircmetros operacionais

como a altura do leito a velocidade superficial do fluido e a concentraccedilatildeo de entrada do

efluente Na praacutetica pretende-se obter a maacutexima fraccedilatildeo de leito saturado e prever o

comportamento dinacircmico do processo de adsorccedilatildeo Um modelo de natureza mecaniacutestica foi

desenvolvido de modo a prever a resposta do sistema face a variaccedilotildees dos paracircmetros

operacionais Foram incluiacutedas as relaccedilotildees de equiliacutebrio bem como o modelo Linear Driving

Force (LDF) para expressar o transporte de massa intraparticular Deste modo verificou-se que

para uma concentraccedilatildeo de entrada fixa em 100 mgL a maacutexima fraccedilatildeo de leito saturado obtida

eacute de 065 para o caso onde se utilizada uma altura de leiro de 65 cm e uma velocidade

superficial de 1443times10-2 cms Foi tambeacutem estudado a influecircncia dos mesmos paracircmetros de

modo a minimizar a diferenccedila entre o tempo de exaustatildeo onde as condiccedilotildees oacutetimas satildeo uma

altura de leito de 25 cm e uma velocidade superficial de 1443times10-2 cms

Para aleacutem disto desenho fatorial do tipo Box-Benken-Design foi utilizado de modo a

determinar quais as variaacuteveis que mais influenciam a eficiecircncia do processo de adsorccedilatildeo em

leito fixo Utilizando uma regressatildeo linear muacuteltipla dois modelos empiacutericos foram construiacutedos

onde foram identificas as condiccedilotildees oacutetimas de operaccedilatildeo de modo a maximizar a fraccedilatildeo de leito

saturado Por fim a regeneraccedilatildeo do adsorvente foi feita utilizando dois agentes quiacutemicos (aacutecido

niacutetrico e NN-Bis(carboxymethyl)-DL-alanine trisodium salt)

Assim deste estudo foi possiacutevel concluir que o adsorvente testado pode ser utilizado em

processos com colunas de leito fixo para a remoccedilatildeo de croacutemio de efluentes liacutequidos

Palavras chave Adsorccedilatildeo croacutemio coluna de leito fixo casca de pinheiro modificada modelo

mecaniacutestico Box-Benken Design regeneraccedilatildeo

vii

INDEX

Acknowledgements i

Abstract iii

Resumo v

List of Figures ix

List of Tables xi

Acronyms xv

1 Introduction 1

Work motivation and scope 1

Thesis objective 1

Thesis structure 2

2 Theoretical background 3

21 Heavy metals as contaminants for the environment 3

22 Methods for removing heavy metals from wastewaters 5

23 Adsorption process 7

24 Factors affecting adsorption processes 8

25 Desorption and regeneration of adsorbents 9

26 Adsorbents 11

261 Properties of the adsorbent 11

262 Low-cost adsorbents 11

27 Fixed bed adsorption 15

271 The breakthrough concept 16

272 Mathematical models of fixed bed adsorption 19

3 State of the art 27

31 Heavy metal removal in fixed bed column 27

32 Mathematical models for fixed bed adsorption 29

4 Materials and methods 33

41 Materials 33

411 Adsorbate 33

412 Adsorbent 33

viii

42 Experimental methods 34

421 Column experiment 34

43 Analytical techniques 35

431 Dynamic Light Scattering (DLS) 35

432 Moisture of the adsorbent 35

433 Bed porosity analysis 35

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF) 36

44 Experimental planning 36

5 Results and discussion 37

51 Experimental and calculated breakthrough curves 37

511 Effect of fluid superficial velocity on breakthrough curves 37

512 Effect of bed depth on breakthrough curves 40

513 Effect of feed concentration on breakthrough curves 41

52 Comparison of the simulation model with the experimental results 43

53 Statistical treatment (DOE) 45

54 Regeneration test 50

55 Validation of the empirical models 53

6 Conclusion and future work 57

REFERENCES 59

7 Appendixes I

71 Appendix A- Particle size distribution I

72 Appendix B- Experimental data II

ix

LIST OF FIGURES

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12) 4

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6

M- (adapted from11) 5

Figure 23 Stages of the adsorption process- (adapted from20 ) 7

Figure 24 Concentration and temperature profiles in adsorption and desorption processes-

(adapted from18) 10

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29) 12

Figure 26 Molecular structure of lignin- (adapted from30) 13

Figure 27 Molecular structure of cellulose- (adapted from31) 13

Figure 28 Pine bark from (Pinus pinaster) 15

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of

operation ndash (adapted from42) 16

Figure 210 Typical representation of a breakthrough curve 17

Figure 211 Breakthrough curve for plug flow in the bed 19

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

21

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature =

25C and pH 5)- (adapted from44) 23

Figure 41 Fixed-bed column apparels used in experiments 34

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

superficial velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm 38

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

bed depths (a) 119906119900=962times10-2 cms (b) 119906119900=1443times10-2 cms and (c) 119906119900=1924times10-2 cms 40

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III)

adsorption in MPB using different feed concentrations 42

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25

cm (b) H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough

curves (982 202 and 302 mgL) 44

Figure 55 Pareto chart of the factors affecting FLS 45

Figure 56 Observed vs predicted values 46

Figure 57 Three-dimensional response surface effect of superficial velocity and height on

the fraction of saturated bed (FLS) 47

Figure 58 Pareto chart of the factors affecting Δtads 48

x

Figure 59 Observed vs predicted values 49

Figure 510 Three-dimensional response surface effect of superficial velocity and height on

Δtads 50

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by

the simulated model 51

Figure 512 Desorption curves 52

xi

LIST OF TABLES

Table 21 The maximum concentration limits for hazardous heavy metals in industrial

effluents- (adapted from8) 3

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted

from16) 6

Table 31 Column studies on adsorption of heavy metals 28

Table 32 General dynamic models for column adsorption process 30

Table 33 More detailed general dynamic models for column adsorption studies 30

Table 34 Additional models to express adsorbate-adsorbent interactions 31

Table 35 Empirical models for column adsorption studies 32

Table 41 Physical properties of MPB 33

Table 42 Parameters and factor levels 36

Table 51 Factorial design of the experiences performed 37

Table 52 Experimental conditions and calculated parameters to assess the process

performance 39

Table 53 Experimental conditions and calculated parameters to assess the process

performance 43

Table 54 Properties and parameter values used in the simulation 44

Table 55 Experimental results vs predicted results 47

Table 56 Experimental results vs predicted results 49

Table 57 Experimental results obtained from the saturation of the column 51

Table 58 Adsorbed and desorbed amount of chromium 52

Table 59 Experimental values vs predicted values for the model built for the case in which

the FLS was considered as dependent variable 54

Table 510 Experimental values vs predicted values for the model built for the case in which

the (tE-tbp) was considered as dependent variable 54

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 3: Sorption of Cr(III) in fixed-bed column by modified pine ...

i

ACKNOWLEDGEMENTS

The time has come one more step completed It is time to embrace the future enter in a

new world of discoveries and challenges Its time to look into the past with a smile on my face

remembering all the moments Ive pass through

First of all I want to thank my parents and my grandmother for all the support help

and encouragement along this path without them it would not be possible To my uncles to my

godmother Joana and to my grandparents also a sincere gratitude

To my supervisors Professor Doctor Liciacutenio Ferreira and Professor Doctor Margarida

Quina my deepest acknowledgment for all the support availability and guidance in all the

moments during this project

To Aline for her kindness availability cooperation guidance and advice during this

semester at the laboratory and also outside of it

To my university family Rita Xana Joana Priscila Vera Miguel Rodrigo Dani and

Joli who have been always there for all the moments and for their friendship Without them it

would be much harder

To all my friends of Mala who always believed in me for their friendship and all the

moments that wersquove passed

To all the Professors of the Chemical Engineering Department of the University of

Coimbra who provided me all the knowledge necessary to complete this course

To all of you my deepest and sincere acknowledgment

iii

ABSTRACT

The adsorption process in fixed bed columns is one of the most advantageous processes

for the removal of heavy metals allowing the treatment of large volumes of effluent The fact

that the bed can be regenerated and the possibility of being reused allowing the recover the

adsorbed metal are other factors in favor of its use in large-scale industrial processes Chromium

is the heavy metal selected in this study because of its relevance in the electroplating coating

and tanning industries The effluents from these industries may contain detrimental

concentrations of Cr

This work aims to analyse the dynamic behavior of a fixed bed column in which the

adsorbent was prepared in the lab Following the previous work mercerized pine bark (MPB)

will be used as adsorbent to remove Cr (III) and to purify a synthetic effluent Thus operational

parameters such as bed height fluid velocity and effluent inlet concentration were tested to

obtain the maximum saturated bed fraction and to predict the behavior of the adsorption process

for each case A mechanistic model was developed for predicting the behaviour of the fixed

bed that includes equilibrium parameters and the Linear Driving Force (LDF) approach to

express the intraparticular mass transport For an inlet concentration of 100 mgL the maximum

saturated bed fraction was 065 obtained for the case where a height of 65 cm and a superficial

velocity of 1443times10-2 cms is used Moreover the influence of the same parameters was also

studied to minimize the difference between the exhaust time and the break time of the column

In this case using an inlet concentration of 100 mgL a bed height of 25 cm and a surface

speed of 1443times10-2 cms were the best conditions

Moreover a Box-Benken-Design was used to determine the most influent inlet variables

(factors) on adsorption efficiency By using multiple linear regression empirical models were

developed and the optimal operating conditions to maximize the saturated bed fraction were

identified Finally a regeneration of the adsorbent material was assessed through two chemical

agents (nitric acid and NN-Bis(carboxymethyl)-DL-alanine trisodium salt)

From this study it was possible to conclude that the biosorbent tested can be used in

fixed-bed column process for remove chromium from liquid effluents

Keywords Adsorption chromium fixed-bed column modified pine bark mechanistic model

Box-Benken-Design regeneration

v

RESUMO

O processo de adsorccedilatildeo utilizando colunas de leito fixo mostra-se como um dos

processos mais vantajosos para a remoccedilatildeo de metais pesados de efluentes industriais O facto

de o leito poder ser regenerado haver a possibilidade de ser reutilizado e recuperar o metal

adsorvido satildeo outros fatores a favor para o seu uso O croacutemio por ser utilizado em processos

de galvanoplastia revestimentos e curtumes ocorre frequentemente no meio ambiente devido

aos efluentes provenientes deste tipo de induacutestrias

Neste trabalho seraacute estudado o comportamento dinacircmico de uma coluna de leito fixo

em que o adsorvente utilizado foi obtido no laboratoacuterio a partir de casca de pinheiro No

seguimento de trabalhos anteriores este material seraacute utilizado como adsorvente para remover

o Cr(III) Foram realizadas vaacuterias experiecircncias para testar diversos paracircmetros operacionais

como a altura do leito a velocidade superficial do fluido e a concentraccedilatildeo de entrada do

efluente Na praacutetica pretende-se obter a maacutexima fraccedilatildeo de leito saturado e prever o

comportamento dinacircmico do processo de adsorccedilatildeo Um modelo de natureza mecaniacutestica foi

desenvolvido de modo a prever a resposta do sistema face a variaccedilotildees dos paracircmetros

operacionais Foram incluiacutedas as relaccedilotildees de equiliacutebrio bem como o modelo Linear Driving

Force (LDF) para expressar o transporte de massa intraparticular Deste modo verificou-se que

para uma concentraccedilatildeo de entrada fixa em 100 mgL a maacutexima fraccedilatildeo de leito saturado obtida

eacute de 065 para o caso onde se utilizada uma altura de leiro de 65 cm e uma velocidade

superficial de 1443times10-2 cms Foi tambeacutem estudado a influecircncia dos mesmos paracircmetros de

modo a minimizar a diferenccedila entre o tempo de exaustatildeo onde as condiccedilotildees oacutetimas satildeo uma

altura de leito de 25 cm e uma velocidade superficial de 1443times10-2 cms

Para aleacutem disto desenho fatorial do tipo Box-Benken-Design foi utilizado de modo a

determinar quais as variaacuteveis que mais influenciam a eficiecircncia do processo de adsorccedilatildeo em

leito fixo Utilizando uma regressatildeo linear muacuteltipla dois modelos empiacutericos foram construiacutedos

onde foram identificas as condiccedilotildees oacutetimas de operaccedilatildeo de modo a maximizar a fraccedilatildeo de leito

saturado Por fim a regeneraccedilatildeo do adsorvente foi feita utilizando dois agentes quiacutemicos (aacutecido

niacutetrico e NN-Bis(carboxymethyl)-DL-alanine trisodium salt)

Assim deste estudo foi possiacutevel concluir que o adsorvente testado pode ser utilizado em

processos com colunas de leito fixo para a remoccedilatildeo de croacutemio de efluentes liacutequidos

Palavras chave Adsorccedilatildeo croacutemio coluna de leito fixo casca de pinheiro modificada modelo

mecaniacutestico Box-Benken Design regeneraccedilatildeo

vii

INDEX

Acknowledgements i

Abstract iii

Resumo v

List of Figures ix

List of Tables xi

Acronyms xv

1 Introduction 1

Work motivation and scope 1

Thesis objective 1

Thesis structure 2

2 Theoretical background 3

21 Heavy metals as contaminants for the environment 3

22 Methods for removing heavy metals from wastewaters 5

23 Adsorption process 7

24 Factors affecting adsorption processes 8

25 Desorption and regeneration of adsorbents 9

26 Adsorbents 11

261 Properties of the adsorbent 11

262 Low-cost adsorbents 11

27 Fixed bed adsorption 15

271 The breakthrough concept 16

272 Mathematical models of fixed bed adsorption 19

3 State of the art 27

31 Heavy metal removal in fixed bed column 27

32 Mathematical models for fixed bed adsorption 29

4 Materials and methods 33

41 Materials 33

411 Adsorbate 33

412 Adsorbent 33

viii

42 Experimental methods 34

421 Column experiment 34

43 Analytical techniques 35

431 Dynamic Light Scattering (DLS) 35

432 Moisture of the adsorbent 35

433 Bed porosity analysis 35

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF) 36

44 Experimental planning 36

5 Results and discussion 37

51 Experimental and calculated breakthrough curves 37

511 Effect of fluid superficial velocity on breakthrough curves 37

512 Effect of bed depth on breakthrough curves 40

513 Effect of feed concentration on breakthrough curves 41

52 Comparison of the simulation model with the experimental results 43

53 Statistical treatment (DOE) 45

54 Regeneration test 50

55 Validation of the empirical models 53

6 Conclusion and future work 57

REFERENCES 59

7 Appendixes I

71 Appendix A- Particle size distribution I

72 Appendix B- Experimental data II

ix

LIST OF FIGURES

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12) 4

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6

M- (adapted from11) 5

Figure 23 Stages of the adsorption process- (adapted from20 ) 7

Figure 24 Concentration and temperature profiles in adsorption and desorption processes-

(adapted from18) 10

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29) 12

Figure 26 Molecular structure of lignin- (adapted from30) 13

Figure 27 Molecular structure of cellulose- (adapted from31) 13

Figure 28 Pine bark from (Pinus pinaster) 15

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of

operation ndash (adapted from42) 16

Figure 210 Typical representation of a breakthrough curve 17

Figure 211 Breakthrough curve for plug flow in the bed 19

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

21

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature =

25C and pH 5)- (adapted from44) 23

Figure 41 Fixed-bed column apparels used in experiments 34

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

superficial velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm 38

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

bed depths (a) 119906119900=962times10-2 cms (b) 119906119900=1443times10-2 cms and (c) 119906119900=1924times10-2 cms 40

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III)

adsorption in MPB using different feed concentrations 42

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25

cm (b) H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough

curves (982 202 and 302 mgL) 44

Figure 55 Pareto chart of the factors affecting FLS 45

Figure 56 Observed vs predicted values 46

Figure 57 Three-dimensional response surface effect of superficial velocity and height on

the fraction of saturated bed (FLS) 47

Figure 58 Pareto chart of the factors affecting Δtads 48

x

Figure 59 Observed vs predicted values 49

Figure 510 Three-dimensional response surface effect of superficial velocity and height on

Δtads 50

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by

the simulated model 51

Figure 512 Desorption curves 52

xi

LIST OF TABLES

Table 21 The maximum concentration limits for hazardous heavy metals in industrial

effluents- (adapted from8) 3

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted

from16) 6

Table 31 Column studies on adsorption of heavy metals 28

Table 32 General dynamic models for column adsorption process 30

Table 33 More detailed general dynamic models for column adsorption studies 30

Table 34 Additional models to express adsorbate-adsorbent interactions 31

Table 35 Empirical models for column adsorption studies 32

Table 41 Physical properties of MPB 33

Table 42 Parameters and factor levels 36

Table 51 Factorial design of the experiences performed 37

Table 52 Experimental conditions and calculated parameters to assess the process

performance 39

Table 53 Experimental conditions and calculated parameters to assess the process

performance 43

Table 54 Properties and parameter values used in the simulation 44

Table 55 Experimental results vs predicted results 47

Table 56 Experimental results vs predicted results 49

Table 57 Experimental results obtained from the saturation of the column 51

Table 58 Adsorbed and desorbed amount of chromium 52

Table 59 Experimental values vs predicted values for the model built for the case in which

the FLS was considered as dependent variable 54

Table 510 Experimental values vs predicted values for the model built for the case in which

the (tE-tbp) was considered as dependent variable 54

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 4: Sorption of Cr(III) in fixed-bed column by modified pine ...

iii

ABSTRACT

The adsorption process in fixed bed columns is one of the most advantageous processes

for the removal of heavy metals allowing the treatment of large volumes of effluent The fact

that the bed can be regenerated and the possibility of being reused allowing the recover the

adsorbed metal are other factors in favor of its use in large-scale industrial processes Chromium

is the heavy metal selected in this study because of its relevance in the electroplating coating

and tanning industries The effluents from these industries may contain detrimental

concentrations of Cr

This work aims to analyse the dynamic behavior of a fixed bed column in which the

adsorbent was prepared in the lab Following the previous work mercerized pine bark (MPB)

will be used as adsorbent to remove Cr (III) and to purify a synthetic effluent Thus operational

parameters such as bed height fluid velocity and effluent inlet concentration were tested to

obtain the maximum saturated bed fraction and to predict the behavior of the adsorption process

for each case A mechanistic model was developed for predicting the behaviour of the fixed

bed that includes equilibrium parameters and the Linear Driving Force (LDF) approach to

express the intraparticular mass transport For an inlet concentration of 100 mgL the maximum

saturated bed fraction was 065 obtained for the case where a height of 65 cm and a superficial

velocity of 1443times10-2 cms is used Moreover the influence of the same parameters was also

studied to minimize the difference between the exhaust time and the break time of the column

In this case using an inlet concentration of 100 mgL a bed height of 25 cm and a surface

speed of 1443times10-2 cms were the best conditions

Moreover a Box-Benken-Design was used to determine the most influent inlet variables

(factors) on adsorption efficiency By using multiple linear regression empirical models were

developed and the optimal operating conditions to maximize the saturated bed fraction were

identified Finally a regeneration of the adsorbent material was assessed through two chemical

agents (nitric acid and NN-Bis(carboxymethyl)-DL-alanine trisodium salt)

From this study it was possible to conclude that the biosorbent tested can be used in

fixed-bed column process for remove chromium from liquid effluents

Keywords Adsorption chromium fixed-bed column modified pine bark mechanistic model

Box-Benken-Design regeneration

v

RESUMO

O processo de adsorccedilatildeo utilizando colunas de leito fixo mostra-se como um dos

processos mais vantajosos para a remoccedilatildeo de metais pesados de efluentes industriais O facto

de o leito poder ser regenerado haver a possibilidade de ser reutilizado e recuperar o metal

adsorvido satildeo outros fatores a favor para o seu uso O croacutemio por ser utilizado em processos

de galvanoplastia revestimentos e curtumes ocorre frequentemente no meio ambiente devido

aos efluentes provenientes deste tipo de induacutestrias

Neste trabalho seraacute estudado o comportamento dinacircmico de uma coluna de leito fixo

em que o adsorvente utilizado foi obtido no laboratoacuterio a partir de casca de pinheiro No

seguimento de trabalhos anteriores este material seraacute utilizado como adsorvente para remover

o Cr(III) Foram realizadas vaacuterias experiecircncias para testar diversos paracircmetros operacionais

como a altura do leito a velocidade superficial do fluido e a concentraccedilatildeo de entrada do

efluente Na praacutetica pretende-se obter a maacutexima fraccedilatildeo de leito saturado e prever o

comportamento dinacircmico do processo de adsorccedilatildeo Um modelo de natureza mecaniacutestica foi

desenvolvido de modo a prever a resposta do sistema face a variaccedilotildees dos paracircmetros

operacionais Foram incluiacutedas as relaccedilotildees de equiliacutebrio bem como o modelo Linear Driving

Force (LDF) para expressar o transporte de massa intraparticular Deste modo verificou-se que

para uma concentraccedilatildeo de entrada fixa em 100 mgL a maacutexima fraccedilatildeo de leito saturado obtida

eacute de 065 para o caso onde se utilizada uma altura de leiro de 65 cm e uma velocidade

superficial de 1443times10-2 cms Foi tambeacutem estudado a influecircncia dos mesmos paracircmetros de

modo a minimizar a diferenccedila entre o tempo de exaustatildeo onde as condiccedilotildees oacutetimas satildeo uma

altura de leito de 25 cm e uma velocidade superficial de 1443times10-2 cms

Para aleacutem disto desenho fatorial do tipo Box-Benken-Design foi utilizado de modo a

determinar quais as variaacuteveis que mais influenciam a eficiecircncia do processo de adsorccedilatildeo em

leito fixo Utilizando uma regressatildeo linear muacuteltipla dois modelos empiacutericos foram construiacutedos

onde foram identificas as condiccedilotildees oacutetimas de operaccedilatildeo de modo a maximizar a fraccedilatildeo de leito

saturado Por fim a regeneraccedilatildeo do adsorvente foi feita utilizando dois agentes quiacutemicos (aacutecido

niacutetrico e NN-Bis(carboxymethyl)-DL-alanine trisodium salt)

Assim deste estudo foi possiacutevel concluir que o adsorvente testado pode ser utilizado em

processos com colunas de leito fixo para a remoccedilatildeo de croacutemio de efluentes liacutequidos

Palavras chave Adsorccedilatildeo croacutemio coluna de leito fixo casca de pinheiro modificada modelo

mecaniacutestico Box-Benken Design regeneraccedilatildeo

vii

INDEX

Acknowledgements i

Abstract iii

Resumo v

List of Figures ix

List of Tables xi

Acronyms xv

1 Introduction 1

Work motivation and scope 1

Thesis objective 1

Thesis structure 2

2 Theoretical background 3

21 Heavy metals as contaminants for the environment 3

22 Methods for removing heavy metals from wastewaters 5

23 Adsorption process 7

24 Factors affecting adsorption processes 8

25 Desorption and regeneration of adsorbents 9

26 Adsorbents 11

261 Properties of the adsorbent 11

262 Low-cost adsorbents 11

27 Fixed bed adsorption 15

271 The breakthrough concept 16

272 Mathematical models of fixed bed adsorption 19

3 State of the art 27

31 Heavy metal removal in fixed bed column 27

32 Mathematical models for fixed bed adsorption 29

4 Materials and methods 33

41 Materials 33

411 Adsorbate 33

412 Adsorbent 33

viii

42 Experimental methods 34

421 Column experiment 34

43 Analytical techniques 35

431 Dynamic Light Scattering (DLS) 35

432 Moisture of the adsorbent 35

433 Bed porosity analysis 35

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF) 36

44 Experimental planning 36

5 Results and discussion 37

51 Experimental and calculated breakthrough curves 37

511 Effect of fluid superficial velocity on breakthrough curves 37

512 Effect of bed depth on breakthrough curves 40

513 Effect of feed concentration on breakthrough curves 41

52 Comparison of the simulation model with the experimental results 43

53 Statistical treatment (DOE) 45

54 Regeneration test 50

55 Validation of the empirical models 53

6 Conclusion and future work 57

REFERENCES 59

7 Appendixes I

71 Appendix A- Particle size distribution I

72 Appendix B- Experimental data II

ix

LIST OF FIGURES

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12) 4

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6

M- (adapted from11) 5

Figure 23 Stages of the adsorption process- (adapted from20 ) 7

Figure 24 Concentration and temperature profiles in adsorption and desorption processes-

(adapted from18) 10

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29) 12

Figure 26 Molecular structure of lignin- (adapted from30) 13

Figure 27 Molecular structure of cellulose- (adapted from31) 13

Figure 28 Pine bark from (Pinus pinaster) 15

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of

operation ndash (adapted from42) 16

Figure 210 Typical representation of a breakthrough curve 17

Figure 211 Breakthrough curve for plug flow in the bed 19

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

21

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature =

25C and pH 5)- (adapted from44) 23

Figure 41 Fixed-bed column apparels used in experiments 34

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

superficial velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm 38

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

bed depths (a) 119906119900=962times10-2 cms (b) 119906119900=1443times10-2 cms and (c) 119906119900=1924times10-2 cms 40

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III)

adsorption in MPB using different feed concentrations 42

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25

cm (b) H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough

curves (982 202 and 302 mgL) 44

Figure 55 Pareto chart of the factors affecting FLS 45

Figure 56 Observed vs predicted values 46

Figure 57 Three-dimensional response surface effect of superficial velocity and height on

the fraction of saturated bed (FLS) 47

Figure 58 Pareto chart of the factors affecting Δtads 48

x

Figure 59 Observed vs predicted values 49

Figure 510 Three-dimensional response surface effect of superficial velocity and height on

Δtads 50

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by

the simulated model 51

Figure 512 Desorption curves 52

xi

LIST OF TABLES

Table 21 The maximum concentration limits for hazardous heavy metals in industrial

effluents- (adapted from8) 3

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted

from16) 6

Table 31 Column studies on adsorption of heavy metals 28

Table 32 General dynamic models for column adsorption process 30

Table 33 More detailed general dynamic models for column adsorption studies 30

Table 34 Additional models to express adsorbate-adsorbent interactions 31

Table 35 Empirical models for column adsorption studies 32

Table 41 Physical properties of MPB 33

Table 42 Parameters and factor levels 36

Table 51 Factorial design of the experiences performed 37

Table 52 Experimental conditions and calculated parameters to assess the process

performance 39

Table 53 Experimental conditions and calculated parameters to assess the process

performance 43

Table 54 Properties and parameter values used in the simulation 44

Table 55 Experimental results vs predicted results 47

Table 56 Experimental results vs predicted results 49

Table 57 Experimental results obtained from the saturation of the column 51

Table 58 Adsorbed and desorbed amount of chromium 52

Table 59 Experimental values vs predicted values for the model built for the case in which

the FLS was considered as dependent variable 54

Table 510 Experimental values vs predicted values for the model built for the case in which

the (tE-tbp) was considered as dependent variable 54

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 5: Sorption of Cr(III) in fixed-bed column by modified pine ...

v

RESUMO

O processo de adsorccedilatildeo utilizando colunas de leito fixo mostra-se como um dos

processos mais vantajosos para a remoccedilatildeo de metais pesados de efluentes industriais O facto

de o leito poder ser regenerado haver a possibilidade de ser reutilizado e recuperar o metal

adsorvido satildeo outros fatores a favor para o seu uso O croacutemio por ser utilizado em processos

de galvanoplastia revestimentos e curtumes ocorre frequentemente no meio ambiente devido

aos efluentes provenientes deste tipo de induacutestrias

Neste trabalho seraacute estudado o comportamento dinacircmico de uma coluna de leito fixo

em que o adsorvente utilizado foi obtido no laboratoacuterio a partir de casca de pinheiro No

seguimento de trabalhos anteriores este material seraacute utilizado como adsorvente para remover

o Cr(III) Foram realizadas vaacuterias experiecircncias para testar diversos paracircmetros operacionais

como a altura do leito a velocidade superficial do fluido e a concentraccedilatildeo de entrada do

efluente Na praacutetica pretende-se obter a maacutexima fraccedilatildeo de leito saturado e prever o

comportamento dinacircmico do processo de adsorccedilatildeo Um modelo de natureza mecaniacutestica foi

desenvolvido de modo a prever a resposta do sistema face a variaccedilotildees dos paracircmetros

operacionais Foram incluiacutedas as relaccedilotildees de equiliacutebrio bem como o modelo Linear Driving

Force (LDF) para expressar o transporte de massa intraparticular Deste modo verificou-se que

para uma concentraccedilatildeo de entrada fixa em 100 mgL a maacutexima fraccedilatildeo de leito saturado obtida

eacute de 065 para o caso onde se utilizada uma altura de leiro de 65 cm e uma velocidade

superficial de 1443times10-2 cms Foi tambeacutem estudado a influecircncia dos mesmos paracircmetros de

modo a minimizar a diferenccedila entre o tempo de exaustatildeo onde as condiccedilotildees oacutetimas satildeo uma

altura de leito de 25 cm e uma velocidade superficial de 1443times10-2 cms

Para aleacutem disto desenho fatorial do tipo Box-Benken-Design foi utilizado de modo a

determinar quais as variaacuteveis que mais influenciam a eficiecircncia do processo de adsorccedilatildeo em

leito fixo Utilizando uma regressatildeo linear muacuteltipla dois modelos empiacutericos foram construiacutedos

onde foram identificas as condiccedilotildees oacutetimas de operaccedilatildeo de modo a maximizar a fraccedilatildeo de leito

saturado Por fim a regeneraccedilatildeo do adsorvente foi feita utilizando dois agentes quiacutemicos (aacutecido

niacutetrico e NN-Bis(carboxymethyl)-DL-alanine trisodium salt)

Assim deste estudo foi possiacutevel concluir que o adsorvente testado pode ser utilizado em

processos com colunas de leito fixo para a remoccedilatildeo de croacutemio de efluentes liacutequidos

Palavras chave Adsorccedilatildeo croacutemio coluna de leito fixo casca de pinheiro modificada modelo

mecaniacutestico Box-Benken Design regeneraccedilatildeo

vii

INDEX

Acknowledgements i

Abstract iii

Resumo v

List of Figures ix

List of Tables xi

Acronyms xv

1 Introduction 1

Work motivation and scope 1

Thesis objective 1

Thesis structure 2

2 Theoretical background 3

21 Heavy metals as contaminants for the environment 3

22 Methods for removing heavy metals from wastewaters 5

23 Adsorption process 7

24 Factors affecting adsorption processes 8

25 Desorption and regeneration of adsorbents 9

26 Adsorbents 11

261 Properties of the adsorbent 11

262 Low-cost adsorbents 11

27 Fixed bed adsorption 15

271 The breakthrough concept 16

272 Mathematical models of fixed bed adsorption 19

3 State of the art 27

31 Heavy metal removal in fixed bed column 27

32 Mathematical models for fixed bed adsorption 29

4 Materials and methods 33

41 Materials 33

411 Adsorbate 33

412 Adsorbent 33

viii

42 Experimental methods 34

421 Column experiment 34

43 Analytical techniques 35

431 Dynamic Light Scattering (DLS) 35

432 Moisture of the adsorbent 35

433 Bed porosity analysis 35

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF) 36

44 Experimental planning 36

5 Results and discussion 37

51 Experimental and calculated breakthrough curves 37

511 Effect of fluid superficial velocity on breakthrough curves 37

512 Effect of bed depth on breakthrough curves 40

513 Effect of feed concentration on breakthrough curves 41

52 Comparison of the simulation model with the experimental results 43

53 Statistical treatment (DOE) 45

54 Regeneration test 50

55 Validation of the empirical models 53

6 Conclusion and future work 57

REFERENCES 59

7 Appendixes I

71 Appendix A- Particle size distribution I

72 Appendix B- Experimental data II

ix

LIST OF FIGURES

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12) 4

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6

M- (adapted from11) 5

Figure 23 Stages of the adsorption process- (adapted from20 ) 7

Figure 24 Concentration and temperature profiles in adsorption and desorption processes-

(adapted from18) 10

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29) 12

Figure 26 Molecular structure of lignin- (adapted from30) 13

Figure 27 Molecular structure of cellulose- (adapted from31) 13

Figure 28 Pine bark from (Pinus pinaster) 15

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of

operation ndash (adapted from42) 16

Figure 210 Typical representation of a breakthrough curve 17

Figure 211 Breakthrough curve for plug flow in the bed 19

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

21

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature =

25C and pH 5)- (adapted from44) 23

Figure 41 Fixed-bed column apparels used in experiments 34

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

superficial velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm 38

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

bed depths (a) 119906119900=962times10-2 cms (b) 119906119900=1443times10-2 cms and (c) 119906119900=1924times10-2 cms 40

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III)

adsorption in MPB using different feed concentrations 42

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25

cm (b) H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough

curves (982 202 and 302 mgL) 44

Figure 55 Pareto chart of the factors affecting FLS 45

Figure 56 Observed vs predicted values 46

Figure 57 Three-dimensional response surface effect of superficial velocity and height on

the fraction of saturated bed (FLS) 47

Figure 58 Pareto chart of the factors affecting Δtads 48

x

Figure 59 Observed vs predicted values 49

Figure 510 Three-dimensional response surface effect of superficial velocity and height on

Δtads 50

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by

the simulated model 51

Figure 512 Desorption curves 52

xi

LIST OF TABLES

Table 21 The maximum concentration limits for hazardous heavy metals in industrial

effluents- (adapted from8) 3

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted

from16) 6

Table 31 Column studies on adsorption of heavy metals 28

Table 32 General dynamic models for column adsorption process 30

Table 33 More detailed general dynamic models for column adsorption studies 30

Table 34 Additional models to express adsorbate-adsorbent interactions 31

Table 35 Empirical models for column adsorption studies 32

Table 41 Physical properties of MPB 33

Table 42 Parameters and factor levels 36

Table 51 Factorial design of the experiences performed 37

Table 52 Experimental conditions and calculated parameters to assess the process

performance 39

Table 53 Experimental conditions and calculated parameters to assess the process

performance 43

Table 54 Properties and parameter values used in the simulation 44

Table 55 Experimental results vs predicted results 47

Table 56 Experimental results vs predicted results 49

Table 57 Experimental results obtained from the saturation of the column 51

Table 58 Adsorbed and desorbed amount of chromium 52

Table 59 Experimental values vs predicted values for the model built for the case in which

the FLS was considered as dependent variable 54

Table 510 Experimental values vs predicted values for the model built for the case in which

the (tE-tbp) was considered as dependent variable 54

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 6: Sorption of Cr(III) in fixed-bed column by modified pine ...

vii

INDEX

Acknowledgements i

Abstract iii

Resumo v

List of Figures ix

List of Tables xi

Acronyms xv

1 Introduction 1

Work motivation and scope 1

Thesis objective 1

Thesis structure 2

2 Theoretical background 3

21 Heavy metals as contaminants for the environment 3

22 Methods for removing heavy metals from wastewaters 5

23 Adsorption process 7

24 Factors affecting adsorption processes 8

25 Desorption and regeneration of adsorbents 9

26 Adsorbents 11

261 Properties of the adsorbent 11

262 Low-cost adsorbents 11

27 Fixed bed adsorption 15

271 The breakthrough concept 16

272 Mathematical models of fixed bed adsorption 19

3 State of the art 27

31 Heavy metal removal in fixed bed column 27

32 Mathematical models for fixed bed adsorption 29

4 Materials and methods 33

41 Materials 33

411 Adsorbate 33

412 Adsorbent 33

viii

42 Experimental methods 34

421 Column experiment 34

43 Analytical techniques 35

431 Dynamic Light Scattering (DLS) 35

432 Moisture of the adsorbent 35

433 Bed porosity analysis 35

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF) 36

44 Experimental planning 36

5 Results and discussion 37

51 Experimental and calculated breakthrough curves 37

511 Effect of fluid superficial velocity on breakthrough curves 37

512 Effect of bed depth on breakthrough curves 40

513 Effect of feed concentration on breakthrough curves 41

52 Comparison of the simulation model with the experimental results 43

53 Statistical treatment (DOE) 45

54 Regeneration test 50

55 Validation of the empirical models 53

6 Conclusion and future work 57

REFERENCES 59

7 Appendixes I

71 Appendix A- Particle size distribution I

72 Appendix B- Experimental data II

ix

LIST OF FIGURES

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12) 4

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6

M- (adapted from11) 5

Figure 23 Stages of the adsorption process- (adapted from20 ) 7

Figure 24 Concentration and temperature profiles in adsorption and desorption processes-

(adapted from18) 10

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29) 12

Figure 26 Molecular structure of lignin- (adapted from30) 13

Figure 27 Molecular structure of cellulose- (adapted from31) 13

Figure 28 Pine bark from (Pinus pinaster) 15

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of

operation ndash (adapted from42) 16

Figure 210 Typical representation of a breakthrough curve 17

Figure 211 Breakthrough curve for plug flow in the bed 19

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

21

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature =

25C and pH 5)- (adapted from44) 23

Figure 41 Fixed-bed column apparels used in experiments 34

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

superficial velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm 38

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

bed depths (a) 119906119900=962times10-2 cms (b) 119906119900=1443times10-2 cms and (c) 119906119900=1924times10-2 cms 40

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III)

adsorption in MPB using different feed concentrations 42

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25

cm (b) H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough

curves (982 202 and 302 mgL) 44

Figure 55 Pareto chart of the factors affecting FLS 45

Figure 56 Observed vs predicted values 46

Figure 57 Three-dimensional response surface effect of superficial velocity and height on

the fraction of saturated bed (FLS) 47

Figure 58 Pareto chart of the factors affecting Δtads 48

x

Figure 59 Observed vs predicted values 49

Figure 510 Three-dimensional response surface effect of superficial velocity and height on

Δtads 50

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by

the simulated model 51

Figure 512 Desorption curves 52

xi

LIST OF TABLES

Table 21 The maximum concentration limits for hazardous heavy metals in industrial

effluents- (adapted from8) 3

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted

from16) 6

Table 31 Column studies on adsorption of heavy metals 28

Table 32 General dynamic models for column adsorption process 30

Table 33 More detailed general dynamic models for column adsorption studies 30

Table 34 Additional models to express adsorbate-adsorbent interactions 31

Table 35 Empirical models for column adsorption studies 32

Table 41 Physical properties of MPB 33

Table 42 Parameters and factor levels 36

Table 51 Factorial design of the experiences performed 37

Table 52 Experimental conditions and calculated parameters to assess the process

performance 39

Table 53 Experimental conditions and calculated parameters to assess the process

performance 43

Table 54 Properties and parameter values used in the simulation 44

Table 55 Experimental results vs predicted results 47

Table 56 Experimental results vs predicted results 49

Table 57 Experimental results obtained from the saturation of the column 51

Table 58 Adsorbed and desorbed amount of chromium 52

Table 59 Experimental values vs predicted values for the model built for the case in which

the FLS was considered as dependent variable 54

Table 510 Experimental values vs predicted values for the model built for the case in which

the (tE-tbp) was considered as dependent variable 54

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 7: Sorption of Cr(III) in fixed-bed column by modified pine ...

viii

42 Experimental methods 34

421 Column experiment 34

43 Analytical techniques 35

431 Dynamic Light Scattering (DLS) 35

432 Moisture of the adsorbent 35

433 Bed porosity analysis 35

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF) 36

44 Experimental planning 36

5 Results and discussion 37

51 Experimental and calculated breakthrough curves 37

511 Effect of fluid superficial velocity on breakthrough curves 37

512 Effect of bed depth on breakthrough curves 40

513 Effect of feed concentration on breakthrough curves 41

52 Comparison of the simulation model with the experimental results 43

53 Statistical treatment (DOE) 45

54 Regeneration test 50

55 Validation of the empirical models 53

6 Conclusion and future work 57

REFERENCES 59

7 Appendixes I

71 Appendix A- Particle size distribution I

72 Appendix B- Experimental data II

ix

LIST OF FIGURES

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12) 4

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6

M- (adapted from11) 5

Figure 23 Stages of the adsorption process- (adapted from20 ) 7

Figure 24 Concentration and temperature profiles in adsorption and desorption processes-

(adapted from18) 10

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29) 12

Figure 26 Molecular structure of lignin- (adapted from30) 13

Figure 27 Molecular structure of cellulose- (adapted from31) 13

Figure 28 Pine bark from (Pinus pinaster) 15

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of

operation ndash (adapted from42) 16

Figure 210 Typical representation of a breakthrough curve 17

Figure 211 Breakthrough curve for plug flow in the bed 19

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

21

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature =

25C and pH 5)- (adapted from44) 23

Figure 41 Fixed-bed column apparels used in experiments 34

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

superficial velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm 38

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

bed depths (a) 119906119900=962times10-2 cms (b) 119906119900=1443times10-2 cms and (c) 119906119900=1924times10-2 cms 40

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III)

adsorption in MPB using different feed concentrations 42

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25

cm (b) H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough

curves (982 202 and 302 mgL) 44

Figure 55 Pareto chart of the factors affecting FLS 45

Figure 56 Observed vs predicted values 46

Figure 57 Three-dimensional response surface effect of superficial velocity and height on

the fraction of saturated bed (FLS) 47

Figure 58 Pareto chart of the factors affecting Δtads 48

x

Figure 59 Observed vs predicted values 49

Figure 510 Three-dimensional response surface effect of superficial velocity and height on

Δtads 50

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by

the simulated model 51

Figure 512 Desorption curves 52

xi

LIST OF TABLES

Table 21 The maximum concentration limits for hazardous heavy metals in industrial

effluents- (adapted from8) 3

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted

from16) 6

Table 31 Column studies on adsorption of heavy metals 28

Table 32 General dynamic models for column adsorption process 30

Table 33 More detailed general dynamic models for column adsorption studies 30

Table 34 Additional models to express adsorbate-adsorbent interactions 31

Table 35 Empirical models for column adsorption studies 32

Table 41 Physical properties of MPB 33

Table 42 Parameters and factor levels 36

Table 51 Factorial design of the experiences performed 37

Table 52 Experimental conditions and calculated parameters to assess the process

performance 39

Table 53 Experimental conditions and calculated parameters to assess the process

performance 43

Table 54 Properties and parameter values used in the simulation 44

Table 55 Experimental results vs predicted results 47

Table 56 Experimental results vs predicted results 49

Table 57 Experimental results obtained from the saturation of the column 51

Table 58 Adsorbed and desorbed amount of chromium 52

Table 59 Experimental values vs predicted values for the model built for the case in which

the FLS was considered as dependent variable 54

Table 510 Experimental values vs predicted values for the model built for the case in which

the (tE-tbp) was considered as dependent variable 54

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 8: Sorption of Cr(III) in fixed-bed column by modified pine ...

ix

LIST OF FIGURES

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12) 4

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6

M- (adapted from11) 5

Figure 23 Stages of the adsorption process- (adapted from20 ) 7

Figure 24 Concentration and temperature profiles in adsorption and desorption processes-

(adapted from18) 10

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29) 12

Figure 26 Molecular structure of lignin- (adapted from30) 13

Figure 27 Molecular structure of cellulose- (adapted from31) 13

Figure 28 Pine bark from (Pinus pinaster) 15

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of

operation ndash (adapted from42) 16

Figure 210 Typical representation of a breakthrough curve 17

Figure 211 Breakthrough curve for plug flow in the bed 19

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

21

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature =

25C and pH 5)- (adapted from44) 23

Figure 41 Fixed-bed column apparels used in experiments 34

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

superficial velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm 38

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different

bed depths (a) 119906119900=962times10-2 cms (b) 119906119900=1443times10-2 cms and (c) 119906119900=1924times10-2 cms 40

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III)

adsorption in MPB using different feed concentrations 42

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25

cm (b) H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough

curves (982 202 and 302 mgL) 44

Figure 55 Pareto chart of the factors affecting FLS 45

Figure 56 Observed vs predicted values 46

Figure 57 Three-dimensional response surface effect of superficial velocity and height on

the fraction of saturated bed (FLS) 47

Figure 58 Pareto chart of the factors affecting Δtads 48

x

Figure 59 Observed vs predicted values 49

Figure 510 Three-dimensional response surface effect of superficial velocity and height on

Δtads 50

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by

the simulated model 51

Figure 512 Desorption curves 52

xi

LIST OF TABLES

Table 21 The maximum concentration limits for hazardous heavy metals in industrial

effluents- (adapted from8) 3

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted

from16) 6

Table 31 Column studies on adsorption of heavy metals 28

Table 32 General dynamic models for column adsorption process 30

Table 33 More detailed general dynamic models for column adsorption studies 30

Table 34 Additional models to express adsorbate-adsorbent interactions 31

Table 35 Empirical models for column adsorption studies 32

Table 41 Physical properties of MPB 33

Table 42 Parameters and factor levels 36

Table 51 Factorial design of the experiences performed 37

Table 52 Experimental conditions and calculated parameters to assess the process

performance 39

Table 53 Experimental conditions and calculated parameters to assess the process

performance 43

Table 54 Properties and parameter values used in the simulation 44

Table 55 Experimental results vs predicted results 47

Table 56 Experimental results vs predicted results 49

Table 57 Experimental results obtained from the saturation of the column 51

Table 58 Adsorbed and desorbed amount of chromium 52

Table 59 Experimental values vs predicted values for the model built for the case in which

the FLS was considered as dependent variable 54

Table 510 Experimental values vs predicted values for the model built for the case in which

the (tE-tbp) was considered as dependent variable 54

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 9: Sorption of Cr(III) in fixed-bed column by modified pine ...

x

Figure 59 Observed vs predicted values 49

Figure 510 Three-dimensional response surface effect of superficial velocity and height on

Δtads 50

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by

the simulated model 51

Figure 512 Desorption curves 52

xi

LIST OF TABLES

Table 21 The maximum concentration limits for hazardous heavy metals in industrial

effluents- (adapted from8) 3

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted

from16) 6

Table 31 Column studies on adsorption of heavy metals 28

Table 32 General dynamic models for column adsorption process 30

Table 33 More detailed general dynamic models for column adsorption studies 30

Table 34 Additional models to express adsorbate-adsorbent interactions 31

Table 35 Empirical models for column adsorption studies 32

Table 41 Physical properties of MPB 33

Table 42 Parameters and factor levels 36

Table 51 Factorial design of the experiences performed 37

Table 52 Experimental conditions and calculated parameters to assess the process

performance 39

Table 53 Experimental conditions and calculated parameters to assess the process

performance 43

Table 54 Properties and parameter values used in the simulation 44

Table 55 Experimental results vs predicted results 47

Table 56 Experimental results vs predicted results 49

Table 57 Experimental results obtained from the saturation of the column 51

Table 58 Adsorbed and desorbed amount of chromium 52

Table 59 Experimental values vs predicted values for the model built for the case in which

the FLS was considered as dependent variable 54

Table 510 Experimental values vs predicted values for the model built for the case in which

the (tE-tbp) was considered as dependent variable 54

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 10: Sorption of Cr(III) in fixed-bed column by modified pine ...

xi

LIST OF TABLES

Table 21 The maximum concentration limits for hazardous heavy metals in industrial

effluents- (adapted from8) 3

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted

from16) 6

Table 31 Column studies on adsorption of heavy metals 28

Table 32 General dynamic models for column adsorption process 30

Table 33 More detailed general dynamic models for column adsorption studies 30

Table 34 Additional models to express adsorbate-adsorbent interactions 31

Table 35 Empirical models for column adsorption studies 32

Table 41 Physical properties of MPB 33

Table 42 Parameters and factor levels 36

Table 51 Factorial design of the experiences performed 37

Table 52 Experimental conditions and calculated parameters to assess the process

performance 39

Table 53 Experimental conditions and calculated parameters to assess the process

performance 43

Table 54 Properties and parameter values used in the simulation 44

Table 55 Experimental results vs predicted results 47

Table 56 Experimental results vs predicted results 49

Table 57 Experimental results obtained from the saturation of the column 51

Table 58 Adsorbed and desorbed amount of chromium 52

Table 59 Experimental values vs predicted values for the model built for the case in which

the FLS was considered as dependent variable 54

Table 510 Experimental values vs predicted values for the model built for the case in which

the (tE-tbp) was considered as dependent variable 54

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 11: Sorption of Cr(III) in fixed-bed column by modified pine ...

xiii

NOMENCLATURE

119860119904 Surface area (m2)

C Concentration (mgL)

Cb Bulk concentration (mgL)

CE Feed concentration of the adsorbate (mgL)

Cs Surface concentration (mgL)

119863119886119909 Axial dispersion coefficient (m2s)

Dm Molecular diffusivity (m2s)

119863119901119890 Effective pore diffusivity (m2min)

119889119901 Particle diameter (m)

119891ℎ Humidity factor

H Height of the bed (cm)

119869 Mass transfer flux (kgm2s)

Kf Freundlich isotherm constant (mg(1n) L1n g)

KL Langmuir isotherm constant (Lmg)

119896119871119863119865 LDF kinetic constant

119898119886119889119904 Mass of adsorbent (g)

119898119889 Mass of the dried material (g)

119898119894 Mass of initial material (g)

119898119899119886 Not adsorbed mass (mg)

119898119905119900119905119886119897 Total mass introduced (mg)

119873 Flux of mass (gm2s)

n Freundlich isotherm constant

119876 Flow rate (mLmin)

q Adsorption capacity (mgg)

119902lowast Adsorption capacity at the equilibrium (mgg)

qm Maximum adsorptive capacity (mgg)

⟨119902⟩ Average adsorbed phase concentration (molkg)

119877119890 Number of Reynolds

1198772 Determination coefficient

Tb Bulk temperature

TS Surface temperature

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 12: Sorption of Cr(III) in fixed-bed column by modified pine ...

xiv

t Time (min)

119905119887119901 Breakthrough time (min)

119905119864 Exhaustion time (min)

tst Stoichiometric time (min)

119905119891 Total operation time (min)

1199060 Superficial velocity (ms)

119906119894 Interstitial velocity (ms)

119881 Volume (mL)

Z Ion valence

Z Bed length (m)

120576 Bed porosity

120576119901 Particle porosity

120591 Residence time (min)

120585 Capacity factor

120588119901 Particle density (kgm3)

120578 Viscosity (Pa s)

∆119867 Enthalpy (J)

120570 LDF factor

λ Conductance

120591 Tortuosity

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 13: Sorption of Cr(III) in fixed-bed column by modified pine ...

xv

ACRONYMS

BBD Box-Benken design

DLS Dynamic Light Scattering

DOE Design of experiments

EDXRF Elemental Analysis by Energy-Dispersive X-Ray Fluorescence

FLS Saturated bed fraction

LDF Linear Driving Force

MCL Maximum concentration limits (mgL)

MPB Mercerized pine bark

MTZ Mass transference zone

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 14: Sorption of Cr(III) in fixed-bed column by modified pine ...

1

1 INTRODUCTION

Work motivation and scope

Chromium is considered a heavy metal and exists in nature in various oxidation states (di-

tri- penta- and -hexa) however Cr(III) and Cr(VI) are the most stable forms in the aquatic

environment With the increase of the industrial activities the environmental pollution has

increased as well Thus industrial effluents from the metal extraction and fabrication leather

tanning facilities coating factories pigments electroplating may contain high concentrations

of chromium Due to its carcinogenic and mutagenic properties Cr(VI) is considered a highly

toxic metal12

Therefore it is necessary to remove chromium from the environment and to control and

treat the industrial outputs before being discharged into the water bodies

There are several methods and techniques including chemical precipitation reverse osmose

ion exchange evaporation flotation and electrodeposition that can remove chromium from

wastewaters However they have considerable disadvantages such the high reagent and energy

consumption incomplete metal removal generation of sludge or waste that requires special

disposals and the expensive equipment and monitoring system3 Besides this these methods are

ineffective at lower concentrations of metal ions

Recently agricultural by-products and waste materials have received growing attention as

an innovative and economical technology in removing heavy metals instead of the conventional

methods Lignocellulosic materials particularly wood residues have a great sorption capacity

for chromium comparable to the other natural or waste sorbents and they have the advantage

of easily recovering and its cost is very low4

In previous research modified pine bark (MPB) was used as an adsorbent for the removal

of Cr(III) from aqueous solutions in batch experiments The obtained results showed that

modified pine bark has an interesting adsorption capacity for the removal of this heavy metal

However for industrial applications of chromium ion removal a continuous system is

needed In this thesis in a fixed-bed column was investigated for the removal of Cr(III) with

modified pine bark as an adsorbent

Thesis objective

This thesis will be focused on the optimization of sorption of Cr(III) from synthetic effluent

in a fixed-bed column using modified pine bark

2

The adsorbent material used in this experiences were characterized in a previous work So

the main goal will be the study of the operational parameters effect (such as the superficial

velocity the bed depth and the feed concentration) on the adsorption efficiency using a

laboratory scale fixed-bed column Therefore a design of experiments (DOE) will be used to

identify and understand the influence of the parameters in the global system Moreover a

multiple linear regression model will be obtained

For predicting the dynamic response of the fixed-bed column a mechanistic model will be

developed including the Linear Driving Force (LDF) model to account the intraparticular mass

transfer This model will be solved on MATLAB and will allow the comparison of the

experimental results with the simulation predictions

Finally a regeneration test will be carried out aiming at reuse in new cycles of saturation

Thesis structure

The present work is organized in 6 chapters The first chapter includes the introduction

where it is explained the scope of the study the objective and the structure of the dissertation

The second chapter contains the theoretical background and the fundamentals of the

adsorptiondesorption processes adsorbents the concepts and the mathematical models of fixed

bed adsorption In the third chapter is present the state of the art where are referenced general

rate adsorption models and empirical models for dynamic fixed bed columns and a literature

review The experimental procedure and the methodologies used during the experiments are

described in the fourth chapter The experimental results and the models for describing the

fixed-bed adsorption process are present and discussed in the fifth chapter

At the end in the sixth chapter the final conclusions of the present work and some

suggestions for future work are indicated

3

2 THEORETICAL BACKGROUND

21 Heavy metals as contaminants for the environment

Since the beginning of industrialization pollutants have been introduced into natural

environments due to technological difficulties or lack of knowledge about the harmful impact

In particular heavy metals are examples of pollutants that in this case are released into the

ecosystems5 Heavy metals exist in nature mainly in ore minerals However their

hazardousness to the public health and environment are mainly due to anthropogenic sources

and the entry in the food chain

The threat posed by heavy metals to the health of living organisms and plants is worsened

because they are persistent and bioaccumulate contaminants especially on aquatic

environments1

The most common heavy metals are iron (Fe) lead (Pb) copper (Cu) mercury (Hg) zinc

(Zn) nickel (Ni) and chromium (Cr)6 Although some of these heavy metals are indispensable

to life of all living beings even at lower concentrations they may have extremely bad effects

due to their toxicity cancerogenic properties and the ability to cause irritation and

inflammation7

Nowadays discharge limits for hazardous chemicals have been established for effluents

which have to be respected by the industries in order to minimize human and environmental

exposure Table 21 shows the maximum concentration limits (MCL) of heavy metals allowed

to be discharged by the industries and the associated health hazards related to each one reported

by USEPA (United States Environmental Protection Agency)

Table 21 The maximum concentration limits for hazardous heavy metals in industrial effluents-

(adapted from8)

Heavy

metal

Associated health hazards MCL

(mgL)

As Skin manifestations visceral cancers vascular disease 0050

Cd Kidney damage renal disorder human carcinogen 001

Cr Headache diarrhea nausea vomiting carcinogenic 005

Cu Liver damage Wilson disease insomnia 025

Ni Dermatitis nausea chronic asthma coughing human carcinogen 020

Zn Depression lethargy neurological signs and increased thirst 080

Pb Brain damage diseases of the kidneys circulatory and nervous

system

0006

Hg Rheumatoid arthritis diseases of the kidneys circulatory and nervous

system

000003

MCL- Maximum concentration limit

4

Examples of natural processes that bring the heavy metals in the environment are erosion

volcanic eruptions and the weathering of minerals9

Heavy metals from anthropogenic sources such as from chemical industry typically have

a high bioavailability due to their soluble and reactive forms These anthropogenic sources

include the production of alloys batteries production coating processes explosive

manufacturing leather tanning improper stacking of industrial solid waste mining phosphate

fertilizer pesticides photographic materials sewage irrigation smelting steel and

electroplating industries printing pigments textiles and wood preservation910

Chromium specifically is very resistant to corrosion and it can be found in nature in form

the of ferric chromite ore (FeCr2O4) crocoite (PbCrO4) and chrome ochre (Cr2O3)9 Chromium

has many oxidation states but the most common are chromium (III) and chromium (VI)

Due to their different oxidation states this heavy metal possesses different toxicities being

the Cr(VI) much more toxic compared to the other oxidation states1

The form that this metal appears in nature is highly influenced by the pH and oxidation-

reduction potential of the aquatic environment Usually the form Cr(III) appears as stable

complexes or in its cationic form while Cr(VI) appears in anionic form as a chromate (CrO42minus)

or dichromate (Cr2O72minus) or in acid form (1198671198621199031198744minus)11

The speciation of chromium as a function of the pH is represented in Figure 21 Is it

possible to observe that until the pH of 3 the dominant speciation is Cr(III) and it appears as

Cr3+ As the pH increases from 4 to 10 the most common species is (119862119903(119874119867)2+) However

between the pH of 6 to 10 chromium may precipitate as Cr(OH)3 Besides this above pH of

10 chromium becomes soluble again in the form of (Cr(OH)4minus)12

Figure 21 Speciation of chromium (III) at a given pH- (adapted from12)

Mo

lar

frac

tio

n

pH

5

Analyzing the Pourbaix diagram Figure 22 is another method to determine at a given pH

which forms chromium takes place in an aqueous solution This potential-pH diagram can be

interpreted as a phase diagram where the lines represent the frontiers between the stability areas

of the different ionic species

During the dissolution of Cr(III) salts hydrolysis reactions can occur and a reduction of

pH values is normally observed by the following equations

1198621199033+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (1)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)2+ + 119867+ (2)

119862119903(119874119867)2+ + 1198672119874 harr 119862119903(119874119867)3 + 119867+ (3)

An increase of the pH results in a decrease of 1198621199033+ concentration and the species

119862119903(119874119867)2+ becomes more abundant on the solution Plus the increase of the pH reduces the

concentration of 119867+ ions favouring the adsorption of 119862119903(119874119867)2+ by the negatively charged

functional groups presented on the adsorbent 13

22 Methods for removing heavy metals from wastewaters

Removal of heavy metals from industrial effluents can be performed through many

conventional processes including unit operations such chemical precipitation coagulation

activated carbon adsorption foam flotation ion exchange electro-deposition membrane

operations and cementation14

pH

11986211990311987442minus

1198621199033+

1198621199032+

Cr

119862119903(119874119867)3

1198671198621199031198744minus

119862119903(

119874119867

)2+

119864119878

119867119864

119881

Figure 22 Pourbaix diagram for chromium at T=25 ordmC in an aqueous solution [Cr(III)]=10-6 M- (adapted

from11)

6

Among these methods precipitation is the most used due to the economic aspects However

the efficiency of this method can decrease due to other complexing agents in industrial

effluents originating an incomplete process and production of toxic sludge7 In addition after

the treatment the effluent usually contains unacceptably total dissolved solids load15

Moreover this treatment methods have high reagent requirement unpredictable metal ion

removal generation of toxic sludge etc Plus for diluted metal ion concentrations precipitation

is ineffective7

On the contrary adsorption is a simple process and due to low-cost adsorbents it has

become the most effective method for removal of toxic pollutants from wastewaters In fact it

is possible to ensure low cost robustness and versatility15

In Table 22 are presented the main advantages and disadvantages of the most common

physico-chemical processes used to treat the industrial wastewaters

Table 22 Advantages and disadvantages of the main physico-chemical processes- (adapted from16)

Treatment method Advantages Disadvantages

Chemical precipitation -Low capital cost

-Simple operation

-Not metal selective

-Sludge generation

-Extra operational cost for sludge

disposal

Coagulation flocculation -Economically feasible

-Bacterial inactivation

capability

-Production of sludge

-High consumption of chemicals

and transfer of toxic compounds

into solid phase

Flotation Low operational costs

-Metal selective

-Low retention times

-Removal of small particles

-High initial capital cost

-High maintenance and operation

costs

Ion-exchange -High separation selectivity

-Less sludge volume produced

-Regeneration of the metal

-High investments and capital costs

-Pre-treatment of effluent required

-Not available for all heavy metals

Adsorption with new

adsorbents

-Low-cost

-Easy operating conditions

-High metal-binding capacities

-Low selectivity

-Production of waste products

Electrochemical treatments -Rapid process

-High separation selectivity

-No consumption of chemicals

-High investment and high

operation cost

-Production of H2

Membrane filtration -Small space requirement

-Low pressure

-High separation selectivity

-High operational cost due to

membrane fouling

-Limited flow rates

7

23 Adsorption process

Adsorption corresponds to the phenomenon which in the molecules (adsorbate) present in

a liquid or gaseous fluid interact at physical level with a solid surface (adsorbent) placed in

vessels or packed columns It occurs as a result of non-balanced forces on the surface of the

solid that attract the molecules present in the fluid which is in contact with the adsorbent17

The adsorption process is based on transport phenomena and speed of mass transference

The main in the adsorption processes is to purify and separate species present in the liquid or

gaseous phase allowing its recuperation This process is often referred as an economic

alternative in most of the cases1819 The adsorption of a solute present in the fluid phase to an

adsorbent comprehends four steps as indicated Figure 2320

Figure 23 Stages of the adsorption process- (adapted from20 )

Step 1) Occurs the transport of the species involving transport through the liquid solution

to the boundary layer around the adsorbent

Step 2) Comprehends the transport by diffusion of the adsorbate from the boundary layer

to the entrance of the pores denominated by external diffusion

Step 3) Includes the transport inside the pores of the adsorbent by molecular diffusion

through the liquid present inside of the pores and diffusion along the surface of the adsorbent

Step 4) Occurs the binding of the adsorbate on an available adsorbent site Depending on

the saturation degree of the adsorbent this step can be faster or slower

Fluid phase

Solid phase

8

The adsorption processes can be classified as chemical or physical considering the kind of

forces involved17

bull The chemical adsorption is the process in which occurs an effective interaction between the

solid surface and the adsorbed and an irreversible chemical bond is formed From this a

formation of a single layer occur above the solid surface with a release of a considerable

amount of energy is observed

bull The physical adsorption is a reversible phenomenon where it is normally observed the

deposition of one or more than one layer of adsorbate over the adsorbent surface Van der

Walls (dispersion-repulsion) forces are usually involved and the energy released is smaller

compared to the chemical adsorption

24 Factors affecting adsorption processes

The adsorption process may be affected by multiple factors 19 Surface area

Since the adsorption is based on surface interactions the intensity of adsorption is normally

proportional to the specific surface area In other words higher the surface areas increase the

extent of adsorption The surface area of a powdered solid adsorbent depends upon its particle

size Smaller the particle size greater is its surface area

Temperature

The kinetics of adsorption and equilibrium parameters are dependent on temperature As

adsorption normally releases heat (exothermic system) according to the Le-Chatelierrsquos

principle the magnitude of adsorption should decrease with rising of temperature It is also

necessary considering that higher temperatures can damage the adsorbent and increase costs

Pressure

The extent of adsorption is directly proportional to the pressure So an increase in the

pressure of the system increases the extent of adsorption At low temperature the extent of

adsorption increases rapidly with pressure This variable has higher influence for gases

pH

This variable determines the concentration of H+ and OH- groups on the solution which may

interfere with the adsorbate Usually it is necessary to control the pH of the solution in order

to control and avoid the dissociation of the active functional groups of the adsorbent such the

hydroxyl carboxyl and phenolic groups

9

Initial concentration

The adsorption process occurs due to a driving force between the adsorbent and the

adsorbate Thus higher inlet concentrations will lead to a higher driving to adsorption The

concentration is also related to the rate of the adsorption ie higher concentrations lead to

higher ratios of adsorption

Adsorbent dosage

The quantity of adsorbate removed is related to the quantity of adsorbent used Thus as

higher dosage used higher the removal quantity

Adsorbent particle size

Since the adsorption process is strongly dependent on the surface area of the adsorbent

small particles are preferable because of their higher surface area Thus the adsorption process

becomes more efficient when smaller particles are used as adsorbents

In this way the adsorption is a complex process that requires optimization to maximize

efficiency and minimize the costs and the capital invested

25 Desorption and regeneration of adsorbents

The process of adsorption in fixed bed column also includes the stage of desorption to

recover the adsorbent The desorption can be defined as the release of a substance (adsorbate)

from a solid (adsorbent) either from the surface or through the surface The desorption can occur

after the saturation by changing the fluid conditions

Since the adsorbent particles have a finite uptake capacity the solid adsorbent may reach

the saturation At this point the rates of adsorption and desorption are equal and equilibrium is

reached Thus it is necessary to regenerate it or dispose of the adsorbent21

Depending on the application the adsorbent may be discarded after the usage if it has no

more economic value In the most of the cases the disposal of the adsorbents it is not an

economical solution and it is only favored when it is very difficult to regenerate and the

adsorbed component as a low economic value

The regeneration is usually carried out to recover the adsorbed substances allowing the

adsorbent to be reused18 It can be accomplished in situ or ex situ to the adsorption vessel

Generally after the column saturation the adsorbent may be regenerated using suitable eluent

solutions

Taking into account that the desorption is a faster process than the adsorption the

regeneration yields smaller volumes of effluent containing concentrated adsorbate solutions22

This fact allows recovering the adsorbate (ie heavy metals) in a simple and economical way

10

Depending on the nature of the absorbentadsorbent bonds there are several mechanisms to

accomplish desorption20

bull Increase of the temperature in the system

bull Change of chemical conditions eg pH

bull Reduction of the partial pressure

bull Purging with an inert fluid

bull Displacement with a more strongly adsorbing species

Figure 24 represents the solute concentration and temperature profiles for a porous

adsorbent particle in contact with a fluid for the cases of the adsorption and desorption

respectively

where TS represents the temperature on the surface of the particle and Tb the temperature in the

bulk Cs and Cb correspond to the concentration of a certain substance in the surface of the

particle and in the bulk respectively

As it is possible to conclude for the case of the adsorption the concentration of a certain

substance is lower inside of the adsorbent particle and higher in the bulk fluid As the adsorption

process goes on this difference in the concentrations become lower and it reaches the

equilibrium For the case of the desorption the opposite happens ie the concentration of a

certain substance inside the particle is higher and as the process runs it tends to lower until

reaches the equilibrium The temperature in the desorption process is higher on the bulk fluid

since usually the desorption process is more efficient at higher temperatures thus the particle

surface should be at a lower temperature than the bulk fluid

Adsorption Desorption

Bulk

fluid

Bulk

fluid

Adsorbent

particles

Figure 24 Concentration and temperature profiles in adsorption and desorption processes- (adapted

from18)

11

26 Adsorbents

261 Properties of the adsorbent

Adsorbents are usually solid particles used in the adsorption processes and their physical-

chemical nature is an important factor in the adsorption since that the capacity and the velocity

of the adsorption depends on the specific surface area porosity specific volume of porous

distribution of the size of porous and the functional groups present on the surface of the

material20 Typically adsorbents have a large range of chemical forms size of porous and

different surface structure23

At industrial level a good adsorbent must18

bull Have high uptake capacity to minimize the quantity of adsorbent used

bull Possess high selectivity to remove only the desired component(s)

bull Have chemical and thermic stability

bull Favor the mass transport for a fast adsorption

bull Have low solubility in contact with the fluid to preserve the quantity of adsorbent and

their chemical and physical properties

bull Have high toughness and mechanic resistance to avoid the smashing and erosion

bull Allow regeneration

bull Have high open porosity

bull Do not promote undesired chemical reactions

bull Be economically viable

262 Low-cost adsorbents

During the last decades several studies developed technologies to remove pollutants from

wastewaters and avoid disposal

Until a few years ago granular activated carbon (GAC) was considered the best adsorbent

for wastewater treatment due to its good capacity of adsorption However there is a need of

replacing it due to its high cost of production and difficulties in regeneration This economic

problem has forced scientists for developing new adsorption materials24

In recent years agricultural waste materials have attracted great interest due to their basic

components including cellulose hemicellulose lignin extractives lipids proteins simple

sugars hydrocarbons and starch containing a variety of functional groups that facilitate

adsorbate complexation25

12

Agricultural waste materials should possess several requirements for the adsorption of

metals such as efficient for the removal of a wide variety of heavy metals maximum removal

of specific metals high adsorption capacity low-processing cost easily regenerateddisposed

off after adsorption and be tolerant for a wide range of concentration of heavy metals26 Each

one of this material has its own specific physical and chemical characteristics such as porosity

surface area and physical strength Adsorption capacities of the adsorbent were also found to

vary depending on their experimental conditions

Nowadays agricultural residues are considered economic and eco-friendly due to their

unique chemical composition and their availability in abundance being a promising option for

the usage as an adsorbent for removal of heavy metals from wastewaters26

Generally lignocellulosic materials including residues of cereals such rice bran rice husk

coconut shells maize corn cob olive or cherry stones peanut shells grapes stalks coffee beans

bark of the trees sugar cane bagasse etc have potential metal sorption capacity which derivate

from their polymers and structure27 The polymers include extractives cellulose

hemicelluloses pectin lignin and protein which facilitates metal complexation helping the

removal of heavy metals This type of materials is used as adsorbents for a wide range of solutes

particularly divalent metal cations

More specifically pine bark for presenting all those characteristics is shown as a material

with the potential to be applied in the removal of heavy metals However because of its

complex structure and depending on the region and the environmental conditions the chemical

composition of the pine bark may change28

Despite all these differences the pine bark is mainly constituted by lignin polysaccharides

such the cellulose and hemicellulose and a low percentage of ash29 Figure 25 shows an

approximate percentage of the chemical composition of the bark of pinus pinaster

Figure 25 Chemical composition of the pinus pinaster bark- (adapted from29)

Extratives 11

Lignin44

Polysaccharides (Cellulose and Hemicellulose)

42

Ash3

13

Lignin is the major component present in the pine bark and it is considered a highly

branched and amorphous phenolic polymer as it is possible to observe in Figure 26 It has a

high surface area and contains various functional groups present in its structure while cellulose

molecules present an ordered structure composed of multiple hydroxyl carbonyl and methoxyl

groups easily accessible making it a promising adsorbent material2830

The polysaccharides such the cellulose and hemicellulose possess an ordered structure as

it is represented in Figure 27 There are four main types of cellulose depending on their

molecular structure or if it was subjected to any kind of chemical or physical treatment3132 The

cellulose I is the most common in nature and the cellulose II is originated by the mercerization

of the cellulose I3334 The cellulose III and IV are obtained also by cellulose I and II by another

kind of treatments such chemical modifications or heating 3435

The hemicellulose is mainly constituted by repetitions of different monosaccharide units

such the glucose and xylose with short branches of sugar units attached giving an amorphous

Figure 26 Molecular structure of lignin- (adapted from30)

Figure 27 Molecular structure of cellulose- (adapted from31)

14

structure Because of it the hydroxyl functional groups become more accessible and react more

easily with metal ions providing a good adsorption rate2831

The extractives representing about 11 of the pine bark composition are formed by a

secondary metabolism of the tree mechanical damage or attack by fungi or insects They are

constituted generally by phenols resins fats and waxes varying from species to species and

their geographic location35

Finally the ash represents the inorganic constituents of the pine bark such as Na K Ca

Mg Zn and P absorbed by the roots during the lifetime of the tree The presence of these

elements on the pine bark affects directly the pH of the material surface with influence on the

adsorption process 28

Despite this the uptake capacity of the most of the lignocellulosic materials is not enough

to achieve a good efficiency their physical stability is very variable and the behavior of the

cellulose as a substrate is highly dependent upon the crystallinity specific surface area and

degree of polymerization16 Thus raw lignocellulosic materials have been modified by various

methods These treatments increase the sorption capacity due to the number of active sites

allowing an easier accessibility to hydroxyl groups turning the lignocellulosic structure of the

material more crystalline Moreover it also provides resistance to microbiological attack and

thermal resistance36 Thereby metal ions adsorption take place on chemical functional groups

such as carboxyl or phenolic (the chemical reactivity is a function of the high reactivity of the

-OH groups present) making possible its removal from wastewaters 37

Considering that at pH 5 the speciation of Cr(III) metal ions is mainly Cr(OH)2+ and the

MPB has many carboxylate groups (COO-) the binding of metal ions on the MPB surface can

be represented by the Eq (4)38

2(minus119862119874119874119867) + 119862119903(119874119867)2+ harr (minus119862119874119874)2119862119903(119874119867) + 2119867+ (4)

where the group (-COOH) corresponds to functional groups present on the surface of the

adsorbent It is also important to work at higher values of pH since the deprotonation increase

thus there are less H+ ions to compete with Cr(OH)2+ ions being possible to retain a greater

amount of metal ions in the adsorbent obtaining a more effective removal However higher

values of pH above 6 can cause the precipitation of the chromium

-Pine bark from (Pinus pinaster)

Nowadays pine bark is a sub product especially from the industry of paper and it is used

for garden decoration or it can also be used as a bio-combustible

15

As seen previously pine bark has all the requirements for its usage as a low-cost adsorbent

specifically for the removal of heavy metals

However its adsorption capacity in raw form is not high enough to achieve good

efficiencies So in order to achieve the best results and properties for the adsorption the raw

pine bark was subjected to an alkali treatment known as mercerization using an alkali

solution39 Mercerization increases the specific surface area of the material turning hydroxyl

groups more accessible due to the change in the amorphous structure to a crystalline providing

a higher degree of substitution40

Figure 28 Pine bark from (Pinus pinaster)

Pinus Pinaster is abundant in Portugal and to valorize the bark a low-cost adsorbent will

be developed for the treatment of polluted effluents with heavy metals

27 Fixed bed adsorption

Industrial processes that use adsorption as a separation process are classified according to

their operation mode The most applied mode of operation is a column packed with adsorbent

particles (fixed bed) through which the effluent will pass adsorbing or removing one or more

components present in it18 After a certain contact time the column will reach saturation and

will not continue to retain the components of interest At this point it is necessary to change

the adsorbent or regenerate it18

The fixed bed column in the adsorption process has several advantages in terms of

assembly cost ease of operation change and facility of changing the operation parameters

easy scale-up and the possibility of regenerating the adsorbent1841

The operation of fixed bed column includes a cycle of saturation or loading the stage of

desorption or regeneration and washing41

16

271 The breakthrough concept

One of the principal requirement to design an adsorption process is the choice of the material

used as adsorbent and adsorbate It is possible to estimate the dynamic or breakthrough capacity

of the column by monitoring the outflow concentration19 Thus it is possible to produce a

breakthrough curve In addition to this the height and width of the column are also important

parameters to describe the system and the operation of the column4142

A characteristic scheme of a breakthrough curve is presented in Figure 29

Figure 29 shows a case where a downward flow is introduced into the column in which

the adsorbent is free of adsorbate As the adsorbate is introduced in the column over time close

to the entrance of the column it is formed a zone known as mass transference zone (MTZ) It

moves along the column until the saturation

In the first stage of the operation (1) the solute is adsorbed in the top of the column and the

effluent of the column is free of adsorbate As the operation continues there will be a time that

almost half of the bed is saturated and a saturated zone is formed (black area) however the

concentration of the adsorbate in the effluent is zero (2) When the MTZ reach the bottom of

the column the breakthrough point (119905119887119901) is attained (3) Usually the breakthrough point is

considered as 1 of the inlet concentration in the column41 Generally the breakthrough time

Figure 29 Breakthrough curve diagram where (1) (2) (3) and (4) are different moments of operation

ndash (adapted from42)

119862119864 119862119864 119862119864 119862119864

t 119905119887119901 119905119904119905 119905119864

119862119862

119864

(1) (2) (3) (4)

17

(119905119887119901) decreases with the increase of the particle size higher concentrations of the solute in the

effluent higher values of pH of the influent high flow rates and low bed depths43

After the system reaches this point the concentration of the solute on the effluent increases

rapidly and the column can still operate until the MTZ achieve the exit of the bed and the

concentration of the effluent reaches 95 of the inlet concentration (4) This point is

denominated the exhaustion time (119905119864) From this point on the concentration of the effluent

increases slowly until equalize the inlet concentration of the solution At this moment the

adsorption process is finished 41

Another important parameter is the stoichiometric time (tst) that is defined as the time where

the area above the breakthrough curve is divided into two equal areas which correspond to the

50 of the bed saturation Through the mass balance at the column it is possible to obtain

119876119862119864119905119904119905 = 120576119881119862119864 + (1 minus 120576)120576119901119881119862119864 119905119904119905 = 120591 (1 + 120585) (5)

where Q is the flow rate (119898119871119898119894119899) 119862119864(119898119892119871) represents the feed concentration 120576 the bed

porosity 119881(1198983) the bed volume 120576119901 the porosity of the particles the 120591(119898119894119899) is defined as

residence time calculated by Eq(6) and 120585 corresponds to the capacity factor calculated by

Eq(7)

120591 = 120576119881

119876

(6)

120585 =(1 minus 120576)

120576

119902

119862119864+

(1 minus 120576)

120576120576119901

(7)

where q (119898119892119892) represents the adsorption capacity

The useful capacity of the bed defined as the instant where the solute is detected in the

effluent of the column corresponds to the breakthrough point and it can be represented as the

S1 area observed in the Figure 210

Figure 210 Typical representation of a breakthrough curve

119905119887119901 119905119864

95119862119864

119905

1119862119864

119862119862

119864

18

The stoichiometric time 119905119904119905 is defined as the time taken for the actual capacity to be used

and it is also the point that breaks the MTZ into equal areas and can be calculated as

119905119904119905 = int (1 minus119862

119862119864) 119889119905

119905infin

0

(8)

where C (119898119892119871) is the concentration of the effluent during the operating period

The saturation bed fraction (FLS) can be calculated as

119865119871119878 =int (1 minus

119862119862119864

) 119889119905119905119887119901

0

int (1 minus119862119862119864

) 119889119905infin

0

=1198781

1198781 + 1198783

(9)

and represents the relationship between the total mass of chromium adsorbed in the column

until the breakthrough point and the mass of chromium introduced into the column through the

feed It can also be determined as a function of the S1 area divided by the total area above the

breakthrough curve until the exhaustion time S3 as represented in the Figure 210

The amount of metal ions 119898119905119900119905119886119897(119898119892) feed to the column can be calculated as

119898119905119900119905119886119897 = 119876times119862119864times119905119891 (10)

Through the breakthrough curve is possible to calculate the total of mass not adsorbed in

mg of the metal ion which corresponds to the area under it For a given flow rate and feed

concentration the total not adsorbed quantity mna (119898119892) can be determined by the following

expression

119898119899119886 = 119876times int 119862 119889119905119905119891

0

(11)

The quantity of metal adsorbed at the equilibrium or the adsorption capacity q (mg metal

g adsorbent) can be calculated from the combination of the previous equations divided by the

amount of adsorbent introduced 119898119886119889119904(119892)

119902 =119898119905119900119905119886119897 minus 119898119899119886

119898119886119889119904 (12)

The shape and time of the breakthrough curve are important parameters that need to be

taken into account for assessing the dynamic behavior of the fixed bed column The inlet

concentration the bed depth the flow rate the particle size of the adsorbent and the adsorption

equilibrium relationships are determinant factors that need to be taken into consideration to

describe this system19

For the cases of plug flow operation the diffusional resistance to mass transfer in the

column is null and if the adsorption process was infinitely fast the system response obtained

19

would be a step set up in t=119905119904119905 as illustrated in Figure 211 However in a real situation it does

not happen so the diffusional mass transfer resistance must be considered

The factors that affect the breakthrough curve can be divided in thermodynamic kinetic and

hydrodynamics The thermodynamic factors are the ones that determine the distribution of the

solute between the fluid and the solid phase which includes the inlet concentration porosity of

the bed and of the particle capacity of the adsorption temperature and pressure These

relationships are described in the adsorption isotherms and were determinate in the previous

work4445

The kinetic factors can be distinguished in four stages of mass transfer the transference

from the solution to the film surrounding the particle the external diffusion (through the film)

the internal diffusion (through the porous structure) and adsorption over the interparticle

surface45 These characteristics are also related with the hydrodynamics of the system that

determines the diffusion mechanisms There are several diffusional mechanisms in the

macropores including the molecular diffusion Knudsen diffusion and surface diffusion The

diffusion on the micro pores is associated with interactions of potential field and steric

effects1945

272 Mathematical models of fixed bed adsorption

A mathematical model is a possible way for describing a system in terms of the relations

between operational variables and parameters In other words a mathematical model tries to

describe the phenomenon occurring in a system to gain prediction capacity46 This kind of

models are used to analyze the behavior of complex systems in situations usually difficult to

observe47

Generally the mathematical modelling is a technique based on mathematical equations to

describe a real system with the aim of predicting his behavior under certain conditions

t 119905119904119905

119862119862

119864

1

Figure 211 Breakthrough curve for plug flow in the bed

20

The project and design of the unit operations such the fixed bed columns is usually

accomplished through the use of mathematical models46-48 The development of mathematical

models and computer simulation are an important tool because it allows the transfer of

technology from the laboratory to industrial scale

In this point of view mathematical models show many useful aspects as the possibility of

add or reject certain hypothesis related to complex phenomena show unexpected responses or

predict the behavior of a system under conditions that were not tested yet However the closer

the model to reality the more complex it will be involving a higher number of parameters

resulting in a greater difficulty to obtain the data from the model and its interpretation49

The main goal of mathematical modelling to describe a phenomenon is the possibility of

reducing the costs associated with real experiences increase the speed of data collection and

the development of control systems

Some aspects are necessary for validation and evaluation of a mathematical model which47

bull Should be able to describe the behavior of the of the real system

bull Should be able to support theories or hypotheses that explain the behavior

bull Must be able to predict the behavior namely the effects of changes in the variables of

the system or in its mode of operation

bull Must be able to respond to changes of variables with the same sensitivity presented by

the experimental model50

Mass balance

Thus the construction of a mathematical model to simulate a fixed bed column is not an

easy task The model must take into account the nonlinearity in adsorption equilibrium

isotherms interference effects due to solute competition by adsorbent sites the resistance of

mass transfer between the fluid phase and the solid phase and the hydrodynamic dispersion

phenomena

For modelling the adsorption of Cr(III) in a fixed-bed the following assumptions were

made51

bull At the beginning of the operation the material inserted in the column (MPB) is assumed

to be pure with zero concentration of Cr(II) in the whole column

bull At time t = 0 min the feeding of the column starts from the top to bottom of it

(descending)

bull All sorption parameters remain constant along the column (axial dispersion coefficient

mass transfer coefficient liquid and solid phase as well as kinetic constants)

21

bull Bed packing and porosity are uniforms throughout the column

bull The flow rate is constant and invariant with the column position

bull Hydraulic effects such as flow preferences and clogging are not considered

bull Mixing in the feed reservoir is perfect

bull The temperature is constant in the feed and the bed (thermal effects are not important)

It is important to note that at any point in the column the solute concentration in the liquid

phase is a function of time and position along the column because the system operates in a non-

stationary state relative to solute concentrations in the liquid phase

Thus the model includes the mass balances in the fluid phase ion-exchange equilibrium

isotherms and intraparticular mass transfer governed by a linear driving force (LDF)

The distribution of a component in the fluid and solid phases is obtained through the mass

balance Neglecting the radial dispersion along the column there are only two independent

variables time (t) and length of the column (z) The transient mass balance includes the term

of axial dispersion convection flow term accumulation in the fluid phase and the adsorption

process on the adsorbent particles52 Thus considering a volume element of the bed [119860times∆119911] in

an adsorption column the mass balance is represented by the following equation

[119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119899119905119890119903119904 119894119899 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] minus [119865119897119906119909 119900119891 119898119886119904119904 119905ℎ119886119905

119890119909119894119905119904 119886 119888119890119903119905119886119894119899 119890119897119890119898119890119899119905 119900119891 119907119900119897119906119898119890

] = [119860119888119888119906119898119906119897119886119905119894119900119899 119900119891

119898119886119904119904 119894119899 119905ℎ119890 119907119900119897119906119898119890 119890119897119890119898119890119899119905

]

(13)

∆119911

119911 = 119871

119911 = 0

119862119894

119902119894

119862119894

120576

Figure 212 Volume of control of the adsorption unit in fixed bed column- (adapted from 52)

22

The flux of mass that enters in a certain volume element 119873119911 is given by Eq (14) and the

flux of mass that exits from a certain volume element 119873119911+∆119911 is given by the Eq (15)

119873119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|zt

(14)

119873119911+∆119911 = 120576119860119904 (1199060119862 minus 119863119886119909120597119862

120597119911)|z+∆119911 t (15)

The term of accumulation of the mass in the element of volume is given by the following

expression

119860119904∆119911 (120576120597119862

120597119905+ 120588119901119891ℎ

120597119902

120597119905)| zt (16)

Considering that the interstitial velocity and axial dispersion coefficient are constants along

the column and replacing Eqs (14) (15) and (16) on the (13) the final mass balance can be

presented by

120597119862(119911 119905)

120597119905+ 119906119894

120597119862(119911 119905)

120597119911+

(1 minus 120576)

120576120588119901119891ℎ

120597⟨119902(119911 119905)⟩

120597119905minus 119863119886119909

1205972119862(119911 119905)

1205972119911= 0 (17)

where 119863119886119909 is the axial dispersion coefficient (1198982119904) 119906119894 the interstitial velocity (119898119904) 120576 the

bed porosity z (119898) the axial position 119862(119898119892119871) the concentration of the species Cr(III) in the

bulk fluid phase ⟨119902⟩ (119898119900119897119896119892119886119889119904119900119903119887119890119899119905) the average adsorbed phase concentration in the

adsorbent particles 120588119901 the density of the adsorbent(1198961198921198983) and 119891ℎ its the humidity factor

Eq (16) is a partial differential equation of second order So the first term represents the

accumulation of the component in the fluid phase the second term represents the mass transfer

in the column due to the convective effects the third term represents the accumulation of the

component in the solid phase and the fourth term represents the mass transfer due to the effects

of axial dispersion46

The term of the axial dispersion (Peclet number) may be described by Eq (18) and the

Reynolds number by Eq (19)

119906119900119889119901

119863119886119909= (02 + 0011119877119890048)

(18)

119877119890 =119906119900120588119891119889119901

120578 (19)

in which the term 119906119900(119898119904) is the superficial velocity 119889119901 (119898) the average particle diameter 120588119891

the density (1198961198921198983) and 120578 (119875119886 119904) the viscosity of the solution respectively

23

Adsorption equilibrium isotherm

The equilibrium isotherms adsorption for the chromium represents the adsorption capacity

119902lowast(119898119892119892) as a function of the adsorbate liquid concentration at the equilibrium 119862119890(119898119892119871)

was adjusted by the Langmuir model

119902lowast(119911 119905) =119902119898119870119871119862119890

1 + 119870119871119862119890

(20)

where qm (119898119892119892) is the maximum adsorptive capacity and 119870119871 a parameter which relates to the

adsorption energy

119870119871 = 119870119871infin119890119909119901 (

(minus∆119867)

119877119879) (21)

The Langmuir isotherm is a theoretical model and assumes that the adsorption occurs in

specific sites of the specific surface without interactions between adsorbed molecules and the

adsorption process is limited to a monolayer coverage53

Another common equation to describe the equilibrium is the Freundlich isotherm

119902lowast = 119870119891times1198621198901119899

(22)

where 119870119891 (mg1-(1n)L1ng) is the Freundlich isotherm constant and n is an isotherm constant that

represents the adsorption intensity

Figure 213 shows the equilibrium isotherms and the experimental data performed in batch

mode in previous works44 for the Cr(III) adsorption

Figure 213 Equilibrium isotherms (initial concentration of Cr(III) = 500 mgL temperature = 25C

and pH 5)- (adapted from44)

For the present work the Langmuir isotherm will be used to express the equilibrium since

this model has a better adjustment to the experimental data

Ce (mgL)

0 30 60 90 120 150 180 210 240 270 300

qe (

mgg

)

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Cr

Langmuir

Freundlich

24

Intraparticle mass transfer

In cases where the geometry of the adsorbent particle is not well defined the use of the

linear driving force model (LDF) appears as an alternative for modeling these processes The

LDF approximation introduced by Glueckauf (1955)54 admits that the mean adsorbed phase

concentration rate is proportional to the difference between the adsorbed phase concentration

in equilibrium with the bulk and the mean adsorbed phase concentration as indicated in Eq

(23)53

120597⟨119902(119911 119905)⟩

120597119905= 119896119871119863119865[119902lowast(119911 119905) minus (119902(119911 119905))] (23)

where 119896119871119863119865 is the LDF kinetic constant and 119902lowast(119911 119905) is the adsorbed phase concentration of the

species in equilibrium with the bulk concentration

The LDF model assumes54

bull Equilibrium between concentrations on bulk fluid and on particle surface

bull The mass transfer kinetics occurs infinitely fast meaning that exists an instantaneous

equilibrium between the pore fluid concentration and the adsorbed concentration

The 119896119871119863119865 used in the model was estimated by Eq (24) based on the equivalence of LDF

models for homogeneous and porous particles

119896119871119863119865 =120570119863119901119890

1199031199012120588119891ℎ(

119889119902lowast

119889119862) (24)

where 120570 is the LDF factor which is equal to 15 for spherical particles 119863119901119890(1198982119898119894119899) represents

the effective pore diffusivity rp (119898) the radius of the particle of the adsorbent and (119889119902lowast

119889119862) the

slope of the equilibrium data obtained by the batch experiments44

Dpe was calculated by the following expression

119863119901119890 =120576119901119863119898

120591

(25)

where Dm is the molecular diffusivity of the solute in water 120591 the tortuosity of the adsorbent

and 120576119901 its porosity The molecular diffusivity expressed in (1198982119904) for metal ions can be

estimated by the Nernst equation (26) and depends on the equivalent ionic conductance of the

metal in solution λ and on the valence of the ion Z55

Dm = 2661 times10minus11λ

|Z| (26)

25

Boundary and initial conditions

The Danckwerts boundary conditions were considered for solving the mass balance

equations

-for z=0 119863119886119909120597119862(119911119905)

120597119911= 119906119894(119862(0 119905) minus 119862119864) (27)

-for z= L 120597119862(119911119905)

120597119911= 0

(28)

where 119862119864 represents the feed composition and L is length of the bed

The following initial conditions were applied which simulates the start-up of the fixed-bed

column

- for t=0 119862(119911 0) = 0 119902(119911 0) = 0 (29)

The numerical solution of the partial differential equations (PDE) that describes the mass

transfer in the fluid phase was obtained by using the Finite Difference technique that converted

PDE into a set of ordinary differential equations (ODE)53

A script was implemented in MATLAB to solve those equations coupled with equations

that describe the intraparticle mass transfer

27

3 STATE OF THE ART

31 Heavy metal removal in fixed bed column

Adsorption process by dynamic methods more specific by column systems has shown to

be advantageous over the batch type operations because the rate of adsorption is highly

dependent on the concentration of the solute in the effluent that need be treated

For column operation the adsorbents are continuously in contact with a fresh solution

which means that the concentration in the solution in contact with the adsorbent present in the

column is relatively constant On the other hand for batch adsorption the concentration of

solute in contact with the adsorbent continuously decreases causing a decrease in the

effectiveness26

The adsorption process in fixed bed columns is the most used among continuous systems

due to their advantages However the disadvantage of this method is that it takes a long

operation time to attain equilibrium thus a large solution volume is necessary to saturate the

column56 Recently some studies were made using fixed bed columns to treat wastewaters and

industrial effluents

Agricultural residues have shown interesting adsorption properties namely low-cost

Despite their good properties and potential to be a future solution for the wastewater treatment

containing heavy metals the investigation in this field is recent

In order to compare the adsorption capacities to metal cations of agricultural waste

materials Table 31 was constructed

As it possible to observe the studies were performed with pH in the range 35 to 58 which

means that all the experiments were carried out in an acid environment favoring the presence

of divalent and trivalent metal cations The adsorption capacity of agricultural waste materials

is in the range of 17 to 10598 mgg but for Cr(III) the capacities were in the range 496 to

6812 mgg Some adsorbents were subjected to chemical treatment However sometimes the

adsorption capacity did not have a significant increase

Considering the operational conditions it is possible to conclude that for the most cases

the higher values of adsorption capacity are achieved when it is used higher initial pollutant

concentration keeping the other parameters constant Moreover higher adsorption capacity

was obtained for lower flow rates For the case of the Orange (citrus cinesis) and Pinus

sylvestris sawdust the bed depth does not have an influence on adsorption capacity

28

Table 31 Column studies on adsorption of heavy metals

MB_LDF Mass balance equations applied to an element of volume of the column in the liquid phase and in the solid phase plus LDF

model BDST Bed depth service time model

Operational conditions Results Ref

Adsorbate Adsorbent Model CE

mgL)

pH H

(cm)

D

(cm)

Q

(mLmin)

q

(mgg)

tbp

(min)

tst

(min)

tE

(min)

Cr(III) Hemp

fibers (Alizarin

S)

Thomas

and Yoonndash

Nelson

13 58 7 15 21 689 110 - 300 57

Thomas and

Yoonndash

Nelson

2614 58 7 15 20 1078 80 - 220 57

Olive stone BDST 10 4 134 15 2 496 55 - - 58

BDST 25 4 134 15 2 789 13 - - 58

BDST 50 4 134 15 2 1225 25 - - 58

BDST 75 4 134 15 2 1172 - - - 58

BDST 100 4 134 15 2 1209 - - - 58

Sargassum

filipecircndula

(CaCl2

02M)

MB_LDF 165-

100

3 306 28 6 3351 - - - 59

Orange (citrus

cinesis)

BDST 20 4 255 22 85 1248 1025 - 2922 22

BDST 20 4 34 22 85 1248 1523 - 3426 22

Biomass of

Seaweed Sargassum

MB_LDF 541 35 306 28 2 6656 - - - 13

MB_LDF 525 35 306 28 6 6032 - - - 13

MB_LDF 525 35 306 28 8 6812 - - - 13

Co(II) Natural

hemp fibers

Thomas

model

25 45-

5

7 15 - 1544 - 2159 - 60

Thomas

model

50 45-

5

7 15 - 1801 - 1261 - 60

MB_LDF

3309 35 306 28 6 8147 - - - 61

Cu(II) Marine alga

Sargassum

MB_LDF 662 35 306 28 6 9291 - - - 24

MB_LDF 10216 35 306 28 6 10598 - - - 24

MB_LDF 2024 35 306 28 6 10439 - - - 24

Pb(II) Pinus sylvestris

sawdust

BDST 10 55 65 15 251 57 443 - - 62

BDST 10 55 85 15 251 64 648 - - 62

BDST 10 55 130 15 251 66 1030 - - 62

BDST 3 55 130 15 10 17 220 - - 62

BDST 5 55 130 15 10 16 121 - - 62

BDST 10 55 130 15 10 22 85 - - 62

Ni(II) Waste of

tea

Thomas 100 4 30 2 5 1113 - - - 63

Thomas 100 4 30 2 10 1057 - - - 63

Thomas 100 4 30 2 20 748 - - - 63

Thomas 50 4 30 2 10 731 - - - 63

Thomas 75 4 30 2 10 945 - - - 63

Thomas 200 4 30 2 10 1117 - - - 63

29

Higher initial concentration of pollutant leads to lower breakthrough time (tbp) while higher

bed depths (more quantity of adsorbent on the column) (tbp) For example in the study of

Orange (citrus cinesis) the column needs more time to reach the exhaustion for the higher bed

depth

The adsorption capacity of the agricultural residues generally found to be satisfactory and

considering the advantages of this type of materials and the promising results of our teamwork

it was decided to investigate the performance of the modified Pinus bark for the removal of

Cr(III) from the wastewater using a fixed-bed column

During the study operational parameters such the initial concentration of pollutant the bed

depth and the superficial velocity will be changed in order to optimize them with the final goal

of maximizing the removal of Cr(III)

32 Mathematical models for fixed bed adsorption

Currently a variety of mathematical models have been used to describe and predict the

breakthrough curves of a column adsorption system However there is still a lack of a

comprehensive review of these models

The optimization of the design and the operation conditions of the fixed bed adsorption

system is an important topic to focus on Thus a reliable model should be able to predict

accurately the breakthrough curve and evaluate the effect of each parameter involved in the

adsorption process56 As consequence to simplify the fixed bed adsorption by making

appropriate assumptions based on a certain system several models were proposed based on

different simplifications

Despite several models has been developed to estimate the behavior of the system the oldest

models remain in use not only because of their good prediction of dynamic adsorption but also

because they allowed determining important operational parameters46

In a process with n variables the optimization of them can be done by keeping (n-1)

variables fixed and compare the effect of different values in the remaining variable on the

predicted breakthrough curves by the model However the main limitation of these models and

therefore the optimization of the process becomes difficult because of the complexity and time-

consuming computation simulation46

Some models assume that the rate of intraparticle diffusion is described by Fickrsquos law

Table 32 shows the main general dynamic models used for that purpose46

30

Table 32 General dynamic models for column adsorption process

Model Equation Ref

Pore diffusion model (PDM) 120576

120597119862

120597119905+

120597119902

120597119905=

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

48

Homogeneous surface diffusion

model (HSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903)

64

Pore and surface diffusion model

(PSDM) 120576

120597119862

120597119905+ 120588

120597119902

120597119905=

119863119904

1199032

120597

1205971199032(1199032

120597119902

120597119903) +

120588119863119890119901

1199032

120597

1205971199032(1199032

120597119902

120597119903)

65

In Table 33 are present another two general dynamic models however more complex that

explains with more detail the diffusional phenomenarsquos also based on Fickrsquos law This models

like the PDM HSDM and PSDM describes the dynamic behaviour of the adsorption process

in fixed-bed column but in this cases some assumptions were made in order to describe more

efficiently the interactions between the adsorbent and the ions present in the fluid phase inside

the porous particles of the adsorbent

Table 33 More detailed general dynamic models for column adsorption studies

Modification Assumption Equation Ref

Macropore

diffusion and

micropore

diffusion

Postulating each adsorbent

pellet is composed of a core and

outer region with macropore

-Diffusion equation of macropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119872119886 1199032 120597119902

120597119903) rc lt rlt rp

-Diffusion equation of micropore region 120597119902

120597119905=

1

1199032

120597

1205971199032 ((119863119890)119898119894 1199032 120597119902

120597119903) 0 lt rlt rc

-Mass conservation at the interface

(119863119890)119872119886 (120597119902

120597119903)

119903rarr119903119888+

= (119863119890)119898119894 (120597119902

120597119903)

119903rarr119903119888minus

64

Shrinking

core theory

model

This model used to describe the

intrapellet adsorption is

controlled by pore diffusion

supposing adsorption firstly

occurs at the outer region of the

adsorbent then the mass

transfer zone moves inward

together with the extending of

the saturated outer region and

shrinking of the unloaded core

-The mass transfer flux at the solid-liquid

phase interface

119869 = 41205871199031199012119896119891(119862 minus 119862119890)

-Diffusion in the pore of adsorbent (Fickrsquos

law)

119869 =4120587119862119890119903119901119903119888119863119890119901

119903119901minus119903119888

-The velocity of mass transfer zone 119889119903119888

119889119905= minus

119869

41205871199031198882120588119902119890

-The mean concentration of the adsorbed

solute

119902 = 119902119890 [1 minus (119903119888

119903119901frasl )

3]

66

31

For solving equations shown in Table 32 and 33 it is necessary to deal with specific

mathematical difficulties The equations showed in Table 34 have been proposed to overcome

some of these issues For example the wave propagation theory the constant pattern theory

and LDF model can be used together with general dynamic models indicated in Table 31 and

32 to globally obtain a solution

In Table 34 are represented a list of the most used models in this category

Table 34 Additional models to express adsorbate-adsorbent interactions

Model Equation Ref

Wave propagation theory -Velocity of the lsquoconcentration wave

119906119888 =119906119894

1 +120588120576

(119889119902119889119862

)

where (119889119902

119889119862) is calculated by the corresponding isotherm

119902119890 = 119891(119862119890)

67

Constant pattern theory -Overall mass transfer equation

120588 (119889119902

119889119862) = 120576119870119871119886(119862 minus 119862119890)

68

Linear driving force (LDF) -Mass transfer coefficient to represent the intrapellet

diffusion rate

119889119902

119889119905= 119896119890(119902119904 minus 119902119886)

54

To modify the intraparticle diffusion models empirical models were developed based on

different simplifying hypotheses making them less complex and less time-consuming

Although being models constructed based on empirical relationships such the Thomas model

the Clark model the BDST model and the Yoon-Nelson model they show a good prediction

and they are usually used to describe the behavior of the breakthrough curve as represented in

Table 35

In addition for obtain the dynamic solution the choice of the isotherm will directly affect

the results of the mathematic model Even though researchers have developed various isotherm

models it is difficult to find one that fits well in all the cases Due to the fact of each adsorbent

has his own structure and characteristic interactions between different ions isotherms can

present diverse shapes So is the best way to determine isotherms is by conducting experiments

In the present work a mathematical general dynamic model that includes a mass balance

an adsorption equilibrium isotherm and the LDF model to describe the mass transfer was chosen

to predict the dynamic behavior of the fixed bed adsorption

32

Table 35 Empirical models for column adsorption studies

Model Equation Ref

Thomas model 119862119905

1198620=

1

1 + 119890119909119901 (119896119879ℎ

119876(1199020119909 minus 1198620119881119890119891119891)frasl )

66

Clark model -Mass conservation equation

120588119869 = minus119906119894120576120597119862

120597119911minus 120576

120597119862

120597119905

69

Bohart-Adams model and

bed depth service time

(BDST) model

119897119899 (119862119865

119862minus 1) = 119897119899 [119890119909119901 (119896119861119902119898

119867

119906)] minus 119896119861119862119865119905

70

Yoon-Nelson model 119897119899

119862

119862119865 minus 119862= 119870119884119873119905 minus 11990512119870119884119873

71

Although the LDF model is not the most complete model to explain the behavior

intraparticular mass transfer it can significantly reduce the computational time showing an

acceptable precision

33

4 MATERIALS AND METHODS

41 Materials

411 Adsorbate

In this study the Cr(III) used were of analytical grade and purchased from Alfa Aesar The

solution of chromium(III) with a concentration of 100 200 and 300 mgL were prepared by

dissolving chromium nitrate nonahydrated Cr(NO3)39H2O with ultrapure water The final pH

of the solution was 5 adjusted with a solution of 1M NaOH For the regeneration tests two

regenerative agents were used NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis)

from Sigma-Aldrich and nitric acid (1198671198731198743) from Fluka

412 Adsorbent

The adsorbent used in this study was pine bark species Pinus pinaster collected in North

of Portugal sub-region of Pinhal Litoral in the Tacircmegarsquos valley44

In the first stage the pine bark was washed with distilled water to remove the dirt fungus

and dust After washing it was dried in the oven at 90 ordmC for 24 h to remove water possible

The oven-dried bark was crushed in a Reischt MUHLER model 5657 HAAN and sieved using

JULABO model SW-21C to desired mesh size (0088mm-0149mm) for being treated

4121 Adsorbent modification

-Mercerization Treatment

Then the washed pine bark was put in contact with a NaOH solution 10 ww at 25ordmC

during 16 h with constant stirring with the goal of increasing the adsorption capacity of the

material

To lower the pH of the modified pine bark it was washed and filtered with tap water until

a neutral and constant pH In the final stage ultrapure water was used to remove any other

contaminants such calcium or chloride that could be present in tap water At the end the

material was dried in an oven at 90 ordmC for 24 h and stored in a desiccator

The main properties are shown in Table 41

Table 41 Physical properties of MPB

ρ (Kgm3) ƒ119919 119941119953 (mm) Surface

area (m2g)

Average pore

diameter (Ǻ)

Pore volume

(cm3g)

320-360 095 348x10-2 - 197x10-1 874 16640 004

34

42 Experimental methods

421 Column experiment

The fixed-bed column studies were performed using a laboratory-scale glass column D-

46539 Dinslaken of 15 mm internal diameter and 120 mm height A peristaltic bomb Minipuls3-

Gilson was used to feed the column with the inlet solution at a predefined flow rate

The column allows to adjust the height of the fixed-bed and was in the vertical position

There were two filters one on the top and other on the bottom to avoid any movement of the

adsorbent in the column

The feed of the column was introduced on the top of it as showed in Figure 41 First a low

flow rate of ultrapure water was introduced to ensure the bed becomes completely wet to

prevent a preferential flow After that the flow rate was set to the desired value and the

chromium solution was introduced Samples were collected at the exit of the column to analyze

the concentration of the metal

In the saturation experiments samples were collected until the outlet concentration reached

the inlet concentration All the experiments were carried out at room temperature (20 plusmn 2ᵒC)

For the case of the regeneration test the same procedure was used but the solution

percolated in the column previously saturated was the regenerating agent The experiments

were conducted at 25degC and 50degC by heating the solution before its feeding to the column

Figure 41 Fixed-bed column apparels used in experiments

Peristaltic bomb

Top filter

Bottom filter

Adsorbent

For

recycling

Sample for

analysis

Solution reservoir

35

43 Analytical techniques

431 Dynamic Light Scattering (DLS)

The particle size distribution of the adsorbent was measured by Dynamic Light Scattering

(DLS) from Malvern

The adsorbent was suspended with ultrapure water Then the particles were analyzed

through this equipment which measures the intensity of the scattered light The fluctuations

caused by the Brownian movements contain the information about the particle size distribution

and can be analyzed as a function of time enabling the evaluation of particle size72

Figure A1 showed in Appendix A depicts particle distribution in the range of 0034824 to

0196671mm with an average diameter of 0099046mm

432 Moisture of the adsorbent

The moisture content was measured in duplicate using a crucible washed and dried for 12h

Then about 1 g of material was weighed and was placed in an oven during 24 h at 105 C and

the dried mass was registered

To calculate the moisture of the material Eq (30) was used

119867() = 119898119894 minus 119898119889

119898119894times100

(30)

where 119898119894 (119892) is the initial weight and 119898119889 (119892) is the weight at the end of 24 h of drying

433 Bed porosity analysis

The bed porosity was measured by filling a certain height of the column with kerosene and

the mass introduced was registered Then the column was filled with the particles of adsorbent

filled with kerosene and the weight registered The difference between the mass of the column

before and after filled with the solid can be transformed into volume which corresponds to the

void spaces

The porosity of the fixed-bed was calculated by Eq (31)

120576 =119881119907119900119894119889119904

119881119905119900119905119886119897

(31)

Two experiences were done (2919 and 3209) and the average value corresponds to

3064

36

434 Elemental Analysis by Energy-Dispersive X-Ray Fluorescence (EDXRF)

EDXRF is an analytical technique which allows determining the elemental chemical

composition of the sample providing both quantitative and qualitative information73

This technique can be classified as an atomic emission method and thus can measure the

wavelength and intensity of light emitted by energized atoms in the sample irradiated by a

primary X-ray beam Since each element emits a characteristic radiation it allows determining

the type of elements present in the sample

The equipment used was the Rigakus ART with a range of 0 to 1000 mgL

44 Experimental planning

-Design of experiments

Generally there are many possible factors that can affect a certain system some may be

critical and have more relevance than others which may have a minor effect or none on a

response of the system It is intended to reduce the number of factors usually to a small set

(2-5) in order to understand what are the most relevant factors and how they affect the dynamics

of the system giving them the major focus74

In this case the methodology of Design of Experiments was used with the aim of identify

which factors have the bigger relevance on the fixed-bed adsorption process and also perceive

the interaction between them

The experiments were designed based on Box-Benken Design (BBD) and on Table 42 are

showed the inputs for the (BBD)

Table 42 Parameters and factor levels

Factor Level

Low (-1) Central (0) High (+1)

Superficial velocity (cms) 962x10-2 1443 x10-2 1924 x10-2

Bed-depth (cm) 25 50 75

37

5 RESULTS AND DISCUSSION

51 Experimental and calculated breakthrough curves

The objective of the study was to determine optimal conditions of two operating variables

(fluid superficial velocity and bed depth) of the Cr(III) sorption process in fixed bed by applying

a Box-Benken Design A set of 9 (3n) experiments were performed according to the conditions

given in Table 51 The performance of the process should be evaluated by comparing the

amount of the solute adsorbed and leaving the column in the effluent as well as the time

required for the entire operation Therefore it is desirable to minimize the difference

tads= (tE ndash tbp) reducing thus the unused portion of the column and on the other hand to

maximize the saturation bed fraction (FLS)

Table 52 Factorial design of the experiences performed

Run Bed depth (cm) Superficial velocity (cms)x102

1 25 962 (10)

2 25 1443 (15)

3 25 1924 (20)

4 5 962 (10)

5 5 1443 (15)

6 5 1924 (20)

7 75 962 (10)

8 75 1443 (15)

9 75 1924 ((20)

The values in brackets correspond to the flow rate in mLmin calculated as 119928 = 119958119952 (120645 119941120784120786)

511 Effect of fluid superficial velocity on breakthrough curves

Since the cross sectional area of the column is constant and uniform implicates that the

linear flow rate (superficial velocity) of the fluid through the packed bed it is directly

proportional to the overall volumetric flow rate percolated into the column The fluid superficial

was used as independent variable in the optimization process because its value usually is kept

constant in the case of the scale-up of fixed-bed adsorbers62

The effect of superficial velocity on the Cr(III) adsorption by MPB was studied by varying

the flow rate through the bed depth from 10 to 15 and 20 mLmin and keeping constant the inlet

solute concentration (CE =100 mgL) and the bed depth Figure 51 shows the plots of the

chromium concentration in the effluent at outlet column versus the contact time when the

superficial velocity varied nearly from 01 to 02 cms for bed heights of 25 5 and 75 cm

38

(a) (b)

(c)

Figure 51 Experimental breakthrough curves of Cr(III) adsorption in MPB using different superficial

velocities (a) H = 25 cm (b) H = 50 cm and (c) H = 75 cm

For all the experiments it can be observed that the breakthrough curve becomes steeper and

the adsorbent gets saturated early as the superficial velocity increases This may be explained

due to the fact of at higher flow rates thus higher superficial velocities the contact time

between the metal ions and the adsorbent decreases leading to lower efficiency of mass transfer

of Cr(III) ions from solution to the adsorbent As it is possible to observe in Fig 51 for each

different bed depth used the breakthrough curve emerged faster with the increase of the

superficial velocity

The conditions of the experimental design and the values of relevant parameters calculated

from the breakthrough curves are summarized in Table 52 In this Table it can be seen that as

the superficial velocity is increased the breakthrough time (tbp) and the exhaustion time (tE)

were found to decrease to all the experiments This can be explained because at higher flow

rates the adsorption rate controlled by the intraparticle diffusion tends to decrease such as the

residence time (τ)75

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 7)

01443 cms (Run 8)

01924 cms (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 1)

01443 cms (Run 2)

01924 cms (Run 3)

t (min)

0 30 60 90 120 150 180 210 240 270 300

C (

mg

L)

0

20

40

60

80

100

00962 cms (Run 4)

01443 cms (Run 5)

01924 cms (Run 6)

39

Table 52 Experimental conditions and calculated parameters to assess the process performance

The values of the stoichiometric time (tst) decrease as the superficial velocity increases as

shown in Table 52 Since the tst splits the breakthrough curve into two equal areas it is expected

that the stoichiometric time takes lower values for the cases when the column reaches the

saturation earlier which corresponds for each bed depth to the higher superficial velocity used

As the adsorption rate is mainly controlled by intra-particulate diffusion the adsorption

capacity (q) tends to increase as the superficial velocity decreases because the residence time is

longer thus the intra-particular diffusion becomes more effective Therefore the Cr(III) metal

ions have more time to diffuse to the inside of the adsorbent particles making possible to reach

higher values of adsorption capacities For example for a bed depth of 5 cm and velocity of

962 x10-2 cms the adsorption capacity (q) is 3024 mgg while for the superficial velocities of

1443x10-2 cms and 1924x10-2 cms the values q estimated are of 2954 and 2791 mgg

respectively

It is now worth analyzing the effect of the superficial velocity on the two response variables

(tads and FLS) that characterize the process performance As already mentioned previously

the breakthrough and the exhaustion times tend to decrease as the superficial velocity increases

consequently tads = (tE - tbp) decreases However the saturated bed fraction (FLS) is little

affected by the velocity change

Regarding the capacity factor (120527) this parameter takes the highest value in the Run 3

corresponding to approximately 366 This was expected since this parameter is dependent on

the stoichiometric time and residence time and adsorption capacity So it is expected that the

capacity factor assumes higher values for the experiments where higher adsorption capacities

and stoichiometric times where achieved and also for lower residence times

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

1 962 25 998 2761 14163 4857 3049 038 014 11401 35829

2 1443 25 993 1351 6239 3143 2945 040 009 4889 34778

3 1924 25 993 1027 6847 2479 3096 040 007 5819 36569

4 962 50 995 4975 23703 8085 3024 061 027 18728 29790

5 1443 50 101 3055 16175 5187 2954 058 018 13119 28662

6 1924 50 100 2031 12700 3712 2791 053 014 10669 27342

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

8 1443 75 97 4326 20759 7343 2664 055 027 16431 27065

9 1924 75 988 3362 15328 5699 2809 057 020 11966 28014

40

512 Effect of bed depth on breakthrough curves

Another parameter that influences the breakthrough curve in adsorption processes is the bed

depth This factor is one of the most influential on the breakthrough point that determines the

amount adsorbed in a fixed-bed column Figure 53 shows the breakthrough curves obtained

for Cr(III) ion adsorption for different bed depths (25 5 and 75 cm) keeping constant the

superficial velocity and inlet Cr(III) concentration

(a)

(b)

(c)

Figure 52 Experimental breakthrough curves of Cr(III) adsorption in MPB using different bed depths

(a) 119906119900=962x10-2 cms (b) 119906119900=1443x10-2 cms and (c) 119906119900=1924x10-2 cms

Contrarily to the effect of the superficial velocity increasing the bed depth the breakthrough

time also increase resulting in a higher operating time to take the column to the exhaustion

point For example as shown in Table 52 the tbp increases from 276 to 811 min in the case

of the superficial velocity of 962x10-2 cms when the bed depth varied from 25 to 75 cm

This is due to the fact of as much quantity of adsorbent introduced in the column more binding

t (min)

0 30 60 90 120 150 180 210

C (

mg

L)

0

20

40

60

80

100

H = 25 cm (Run 3)

H = 50 cm (Run 6)

H = 75 cm (Run 9)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

C (

mgL

)

0

20

40

60

80

100

H = 25 cm (Run 1)

H = 50 cm (Run 4)

H = 75 cm (Run 7)

t (min)

0 30 60 90 120 150 180 210 240

C (

mg

L)

0

20

40

60

80

100

H =25 cm (Run 2)

H = 50 cm (Run 5)

H = 75 cm (Run 8)

41

sites are available for sorption resulting in a broadened mass transference zone (MTZ) For

smaller bed depths the axial dispersion phenomena predominate in the mass transfer process

leading to a reduction of the diffusivity of the metal ions Consequently the metal ions do not

have enough time to diffuse into the adsorbent leaving the column earlier62

The contact time () between the solution and adsorbent is higher when the bed depth

increases This results in a higher adsorption capacity because the metal ions had more time to

diffuse deeper inside the adsorbent particles which provide greater surface area Therefore

higher amounts of Cr(III) ions are removed from the solution whose values are obtained from

the area above the breakthrough curves

The adsorption capacity (q) values shown in Table 52 exhibit non-significant changes with

variation of bed height Comparing the q values where superficial velocity was kept constant

it is possible to observe that the adsorption capacity remained practically unalterable This can

also be expected because since the adsorption capacity does not depend on the quantity of

material used being an intrinsic characteristic of the adsorbent For all the experiments the

maximum adsorption capacity was determined as approximately 31 mgg concerning to the

Run 3 which is similar to that found in the batch studies qm = 3141 mgg44

Concerning to the stoichiometric time (tst) this parameter increases with the increase of the

bed depth as can be seen in Table 52 The higher value of tst (124 min) was obtained for the

operating conditions (uo = 962x10-2 cms and H =75 cm) that lead to a higher fraction of used

bed (Run 7) in which FLS = 062 The response variable FLS takes higher values for higher

bed depths since the saturated bed fraction (FLS) is quite influenced by the breakpoint time

(tbp) Moreover the variable (tE - tbp) tends also to increase as the bed depth increases at constant

superficial velocity

In order to maximize FLS observing the results of the experiments carried on it was

possible to conclude that the optimum operating point corresponds to the conditions of Run 7

where the bed depth and superficial velocity are 75 cm and 962x10-2 cms respectively

513 Effect of feed concentration on breakthrough curves

The performance of the adsorption of Cr(III) ions in fixed-bed column was also tested

varying the feed concentration of chromium These experiments were performed maintaining

constant the optimal conditions identified from the experimental design results (uo = 962x10-2

cms and H = 75 cm) and with inlet solute concentrations of 982 202 and 302 mgL

In Figure 53 are displayed the results of the effect of the variation of the inlet concentration

of Cr(III) on the breakthrough curves It can be seen that as the inlet concentration (CE)

42

increases the saturation curve becomes steeper In addition the adsorbent reaches the saturation

faster and the breakthrough time decreases with the increase of the feed concentration

Contrarily for lower inlet concentrations the breakthrough curve appears later and

consequently the adsorbent takes more time to saturate

Figure 53 Effect of the feed concentration on the experimental breakthrough curve of Cr(III) adsorption

in MPB using different feed concentrations

As Table 53 shows the breakthrough time (tbp) the exhaustion time (tE) and the

stoichiometric time (tst) assumes lower values when higher feed concentrations are used For

example a reduction of 621 in tst was verified when CE was tripled This may be explained

by the fact that the available binding sites of the adsorbent become saturated more quickly at

high metal concentration entering into the column Under these conditions there is a higher

concentration gradient which leads to a faster solute transport due to an increase diffusional

coefficient or mass transfer coefficient57 Therefore it can be concluded that the mass

transference zone (MTZ) becomes smaller due to an increase of metal loading rate At lower

solute concentrations meaning lower concentration gradient the driving force is not sufficient

to enhance mass transfer and the MTZ is enlarged This can also be observed by the values of

tads presented in Table 53 The smaller the value of this parameter is the smaller is the mass

transfer zone In other words the lower this value steeper the breakthrough curve is ie closer

of ideal saturation conditions

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360 390

CC

o (

mg

L)

00

02

04

06

08

10

CE

= 982 mgL (Run 7)

CE

= 202 mgL (Run 10)

CE

= 302 mgL (Run 11)

43

Table 53 Experimental conditions and calculated parameters to assess the process performance

Run uotimes102

(cms)

H

(cm)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

7 962 75 982 8112 28741 12397 3036 062 041 20629 30476

10 962 75 202 3919 20434 6206 3126 061 041 16516 16516

11 962 75 302 2512 1638 4694 3535 050 041 13868 13868

Regarding the parameter tads = (tE - tbp) it was found to be the lowest when a concentration

of 302 mgL is used because the adsorbent molecules are subjected to a higher driving force

leading to a faster saturation of the column thus the time between the breakthrough point and

the exhaustion point is significantly reduced

Analysing Table 53 it is also possible to observe that a small increase ( 17) in the

adsorption capacity (q) is noticed when CE varied from 982 to 302 mgL which is consistent

with the trend of the equilibrium isotherm obtained for the system modified pine barkCr(III)

The maximum adsorption capacity was achieved for CE = 302 mgL which corresponds to

3535 mgg The capacity factor (120527) that depends on the ratio qCE tends to increase as the inlet

concentration decreases affecting directly the fraction of saturated bed (FLS) FLS was

approximately doubled when a one-third reduction in CE is verified This means that operating

at lower metal feed concentrations allows to use larger column capacity and maximize the

sorption yield

52 Comparison of the simulation model with the experimental results

To analyze the robustness and precision of the model which comprises a set of equations

defined in section 272 experimental breakthrough curves are shown together in the same plot

with ones obtained from the simulations using the model

In order to solve the model equations it was necessary to use the information displayed in

Table 54 namely the properties of the adsorbent (MPB) the bed characteristics equilibrium

parameters and mass transfer parameters The values considered for the density and diameter

of the particles had to undergo minor adjustments inside the ranges shown in Table 54 due to

the physical heterogeneous nature of the adsorbent material

44

Table 54 Properties and parameter values used in the simulation

Adsorbent (MPB) Bed

properties

Equilibrium

parameters

Mass transfer parameters

ρ (Kgm3) = 320-360 L (mm) = 25-75 119902119898119886119909 (mgg) = 3141 119863119901119890 (cm2min) = 1314x10-4

ƒ119919 = 095 D (mm) = 15 119870119871 (Lmg) = 008 119863119886119909 (cm2min) = 286x10-2 ndash 567x10-2

119889119901 (mm) = 55x10-2 - 99x10-2 ε = 0306 dqdC = 098

In Figure 54 are illustrated the obtained experimental data for each case of study and

ones obtained from the simulations

(a)

(b)

(c)

(d)

Figure 54 Experimental and simulated breakthrough curves of Cr(III) adsorption (a) H = 25 cm (b)

H = 50 cm (c) H = 75 cm and (d) Effect of feed concentration on breakthrough curves (982 202 and

302 mgL)

45

The comparison of the experimental and calculated breakthrough curves shows that the

model satisfactorily fits the experimental data corresponding to the dynamics of the sorption

process of Cr(III) using a column packed with chemically modified pine However as it is

possible to observe for all the cases the model does not describe accurately the behavior of the

system when the column is close to the saturation This may be related to the intraparticular

mass diffusion behavior Since the LDF equation used is a simplified model that does not allow

interpret accurately the concentration profile inside the particle due to intraparticular mass

transfer resistance

To surpass this problem a more complete model should be considered incorporating

differential equations to describe more precisely the mass transfer inside the adsorbent in order

to get a more accurate description of the reality However this model will become more

complex requiring advance numerical methods to solve Even so it is possible to conclude that

the best predictions of the model were achieved for the cases in which lower bed depths (25

and 5 cm) and higher superficial velocity were used as well as higher metal inlet concentration

53 Statistical treatment (DOE)

As already mentioned above the response variables tads and FLS are important parameters

for the optimization of the sorption process in a fixed bed A set of experiments were performed

accordingly to a Box-Benken-design factorial design where the factors considered were the

superficial velocity and the bed height (Table 41)

A first analysis of the experimental breakthrough curves by using the software Statistica7

considering as the dependent variable the saturated bed fraction (FLS) was carried out From

the software it was possible to obtain the Pareto chart shown in Figure 55

Figure 55 Pareto chart of the factors affecting FLS

Pareto Chart of Standardized Effects Variable FLS ()

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

-459251

-918697

-151573

3996578

7431807

p=05

Standardized Effect Estimate (Absolute Value)

Superficial velocity (cms)(Q)

1Lby2L

(2)Superficial velocity (cms)(L)

Height (cm)(Q)

(1)Height (cm)(L)

46

It allows to assess the extent and significance of the effects of the factors on FLS at the 95

confidence level As it can be observed the bed height and the superficial velocity are

significant variables for the column utilization efficiency where the height is the variable that

most influences the response variable FLS

A predictive model relating the FLS to the independent variables (H and uo) was generated

by the software and is given by Eq (32) The comparison between the experimental and model

predicted values can be seen in Fig 56 and Table 55 It is possible to conclude that the values

that exhibit higher deviation from ones predicted by the model are the central ones while those

of lower and higher magnitude are closer to the estimated model values A high value for the

determination coefficient (1198772 = 09613) was achieved

Figure 56 Observed vs predicted values

119865119871119878 = 093 + 019119867 minus 0011198672 minus 1061199060 + 43211990602 minus 0121198671199060

(32)

In Table 55 are presented the experimental results the predicted values the error and the

deviation of the results

Analysing Table 55 it is visible that in the most of the cases the relative error of the results

are lower than 5 In the experiments performed with an inlet concentration of 100 mgL a

bed depth comprehended between 25 and 75 cm and superficial velocities of 962 x10-2 cms

and 1443x10-2 cms FLS is predicted with good accuracy by the model

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=0009486

DV FLS ()

035 040 045 050 055 060 065

Observed Values

035

040

045

050

055

060

065

Pre

dic

ted

Va

lue

s

47

Table 55 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run FLS

experimental

FLS

predicted

Relative

error ()

25

962 1 0384 0400 436

1443 2 0398 0385 311

1924 3 0395 0390 109

5

962 4 0609 0595 234

1443 5 0579 0566 235

1924 6 0529 0557 526

75

962 7 0618 0615 040

1443 8 0546 0572 476

1924 9 0573 0549 411

The surface response depicted in Figure 57 shows that the FLS is higher for higher bed

heights and lower superficial velocity as the Pareto chart suggests Thus analyzing the

response surface plot generated by the predictive model is possible to identify the region of

optimum values for the independent variables leading to maximization of the saturated bed

fraction An analytical optimal solution was found (H = 65 cm and uo = 962x10-2 cms)

using the MATLAB software to maximize FLS described by the model in range of the

tested variables This means that this point corresponds to the global optimum which allows

obtaining the maximum yield in the bed utilization within defined ranges

Figure 57 Three-dimensional response surface effect of superficial velocity and height on the fraction

of saturated bed (FLS)

On the second part the same analyze was carried on with the goal was to minimize the

difference between the exhaustion time and the breakthrough point tads = (tE-tbp) In this case

48

the response variable was tads and the independent variables remained the same as in the

previous study

From the Pareto chart shown in Figure 58 it is possible to observe that the bed height

remains the variable with higher impact on the response variable However the superficial

velocity is also a significant variable which negatively affects tads ie tads will take lower

values when the superficial velocity increase

Figure 58 Pareto chart of the factors affecting tads

The predictive model generated by the software is given by Equation (33) Analysing Fig

59 it is noticeable that there is a good agreement between the experimental and model predicted

values For this case the determination coefficient was of 1198772 = 09769 which suggests that

this model explains more accurately the results than the previous one

Δ119905119886119889119904 = 16819 + 6425119867 minus 3711198672 minus 2601141199060 +times74449811990602 minus 64061198671199060 (33)

Pareto Chart of Standardized Effects Variable dt (min)

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

-116277

-183851

2472403

-687215

8293554

p=05

Standardized Effect Estimate (Absolute Value)

1Lby2L

Superficial velocity (cms)(Q)

Height (cm)(Q)

(2)Superficial velocity (cms)(L)

(1)Height (cm)(L)

49

Figure 59 Observed vs predicted values

Table 56 presents the experimental results the predicted values the error and the deviation

of the results

Table 56 Experimental results vs predicted results

H

(cm)

uotimes102

(cms)

Run tads

real (min)

tads

predicted (min)

Relative

error ()

25

962 1 11401 10891 448

1443 2 4889 6222 2726

1924 3 5820 4997 1413

5

962 4 18728 18464 141

1443 5 13120 13024 073

1924 6 10669 11029 337

75

962 7 20629 21404 376

1443 8 16431 15194 753

1924 9 11966 12429 387

For this case the relative error values shown in Table 56 for some experiments are

relatively high when compared with the previous study (Table 55) even considering the

goodness of fit of the model given by Eq (33) with high R2 value

In this way it is possible to observe that the model provides a better prediction for the

experiments carried out with the bed depths of 75 and 5 cm For those conducted on lower bed

depth (25 cm) the model does not seem to have an accurate prediction since the predicted

values possess a higher relative error associated

Observed vs Predicted Values

2 3-level factors 1 Blocks 9 Runs MS Residual=1755514

DV dt (min)

20 40 60 80 100 120 140 160 180 200 220 240

Observed Values

20

40

60

80

100

120

140

160

180

200

220

240

Pre

dic

ted

Va

lue

s

50

Figure 510 depicts the three-dimensional response surface relationship between superficial

velocity and the bed height on tads This figure shows that the minimization of the difference

between the exhaustion time and the breakthrough point occurs for the lower bed height values

and higher superficial velocity This means that under these conditions the trend of the column

saturation is closer to ideal conditions compared to the experiments where the superficial

velocity used is lower It can also be concluded that the intraparticular diffusional effect has

more importance in the cases where the solution is fed to the column at lower superficial

velocities since the (tE-tbp) increases and the bed saturation takes longer

Figure 510 Three-dimensional response surface effect of superficial velocity and height on tads

54 Regeneration test

When the adsorbent reaches the saturation ie equilibrium conditions are achieved a

regeneration step of the adsorbent may be crucially important to reduce the operation cost and

to open the possibility of recovering the metal removed from the liquid phase by adsorption

In previous batch studies38 the desorption Cr(III) results showed that relatively high

removal efficiencies were obtained using as regenerants the biodegradable complexing agent

NN-Bis(carboxymethyl)-DL-alanine trisodium salt (NN-Bis at 50 C) and nitric acid (HNO3 at

25 C) Therefore N N-Bis and HNO3 and were tested to regenerate the biosorbent preloaded

with Cr(III)

The saturation experiments were performed with a bed depth of 65 cm and superficial

velocities of 962x10-2 cms (Run 12 before desorption with NN-Bis) and 1443x10-2 cms (Run

51

13 before desorption with HNO3) In order to maximize FLS it was used a feed concentration

of 100 mgL Figure 511 shows breakthrough curves corresponding to Runs 12 and 13

Figure 511 Experimental breakthrough curve (Run 12 and Run 13) and the data obtained by the

simulated model

The experimental breakthrough curves depicted in Fig 511 were described by the

mathematical model aforementioned and it is possible to observe that the model fits reasonably

well the experimental values giving a good description of the dynamic response of the system

According to Table 57 the breakpoint time and the exhaustion time are 6093 and 25369 min

for Run 12 and 5278 and 18550 min for Run 13 As might be expected the lowest times are

associated with the experiment conducted on high flow rate (Run 13) For these conditions

unexpectedly higher values of the adsorption capacity (q=3257 mgL) and FLS (065) were

achieved Regarding the q and FLS for Run 12 the obtained values were similar to ones

obtained in other experiments

Table 57 Experimental results obtained from the saturation of the column

Run H

(cm)

uotimes102

(cms)

CE

(mgL)

tbp

(min)

tE

(min)

tst

(min)

q

(mgg)

FLS τ

(min)

tads

(min)

120527

12 65 962 100 6093 25369 9733 3042 062 035 19276 27585

13 65 1443 100 5278 18550 8017 3257 065 023 13272 34101

With respect to regeneration where the main goal is the recovery of metal ions adsorbed in

the saturated biosorbent thus giving the possibility of reusing it reducing the process cost a

possible Cr(III) desorption mechanism using acid as regenerant could be38

(minus119877119862119874119874)2119862119903119874119867 + 3119867+ harr 2(minus119877119862119874119874119867) + 1198621199033+ + 1198672119874 (34)

t (min)

0 30 60 90 120 150 180 210 240 270 300 330 360

C (

mg

L)

0

20

40

60

80

100

Run 12

Run 13

Model

52

By the previous equation it is possible to observe that during the desorption process the

H+ ions take the place of the chromium ions present in the adsorbent thus allowing their release

and the possibility of regenerating the bed The results of the regeneration tests are illustrated

in Figure 512

Figure 512 Desorption curves

In order to determine the efficiency of the desorption it was necessary to know the

amount of chromium adsorbed by the MPB during the saturation step and the amount of Cr(III)

desorbed during the regeneration In Table 58 are present the both values for the regeneration

with NN-Bis and HNO3

Table 58 Adsorbed and desorbed amount of chromium

Run Total mass of chromium adsorbed

119924119938 (mg)

Total mass of chromium desorbed

119924119941 (mg)

12 9734 549

13 10422 4883

Eq (35) was used to calculate the desorption efficiency

119863119890119904119900119903119901119905119894119900119899() =119872119889

119872119886times100

(35)

where Md and Ma corresponds to the total amount of chromium desorbed and adsorbed

(119898119892) respectively Md was calculated from the numerical integration of the regeneration

curves

t (min)

0 5 10 15 20 25 30 35 40 45 50 55 60

C (

mg

L)

0

100

200

300

400

500

600

NN-Bis (Run 12)

HNO3 (Run 13)

53

A very low regeneration efficiency (56) was observed when Cr(III) is eluted with NN-

Bis This result was not expected since in batch studies the efficiency achieved was close to

43 A possible explanation to the lower value of efficiency could be due to the fact of the

regenerative process with this complexing agent to be controlled by a very slow kinetics In

addition the experiment duration (43 min) did not promote enough contact time between the

regenerating agent and the sorbent to achieve an efficient desorption Analysing Figure 512 it

can be noticed that the behavior of the curve indicates that the desorption process was not as

expected due to the absence of the typical lsquopeakrsquo which corresponds to the fast release of the

metal ions present in the adsorbent toward the effluent In this way the behavior of the curve

for this case resembles more a bed washing

The elution curve of chromium ions with (HNO3) allowed determining a regeneration

efficiency of 4685 This value still to be low and is not consistent with the result obtained

from batch studies where almost 82 of desorption was found Figure 512 shows that for this

case there is the occurrence of a peak in the elution curve for Cr(III) where its concentration is

about 62 times higher than the feed concentration (100 mgL) used during the MPB saturation

This means that the regeneration of the bed with nitric acid promoted a fast desorption of the

chromium ions from the adsorbent After the elution to reach 10 min the regeneration process

becomes slower and controlled by dispersive effects ie the curve exhibits a long tail

consequently a long time will be needed to completely regenerate the adsorbent

An increase in the (HNO3) concentration in further desorption processes may be more

effective in the removal of a larger amount of chromium ions from the biosorbent however the

loss of biomass material can be significant For the tested conditions it was possible to calculate

a material loss of 75 that also could have led contributed to a low removal efficiency Thus

the possibility of the presence of another speciation of chromium in MPB for the both cases

may influence their interactions with the eluent solution reducing the efficiency removal

55 Validation of the empirical models

To validate the previous empirical models developed by the software Statistica7 the

experiments carried out to study the effect of the feed concentration (Run 10 and Run 11) as

well as the pre-saturations for the regeneration tests (Run 12 and Run 13) were used The Runs

10 and 11 were performed with optimal values of the bed height and superficial velocity (H=75

cm and uo= 962x10-2 cms) and where the concentration varied from 202 to 302 mgL For the

saturation of the column before the regeneration tests one of the experiments (Run 12) was

54

conducted under optimal conditions of H and uo analytically determined by the regression

model used to maximize FLS and the other experiment (Run 13) was performed using a bed

depth of 65 cm and a superficial velocity of 1443x10-2 cms

Table 59 Experimental values vs predicted values for the model built for the case in which the FLS

was considered as dependent variable

Run CE

(mgL)

FLS

experimental

FLS

predicted

Relative

error ()

10 202 0614 0615 031

11 302 0501 0615 2278

12 100 0616 0628 201

13 100 0645 0591 849

From Table 59 it is possible to conclude that the model provides a good prediction for the

cases where the feed concentration was 202 mgL (Run 10) and when the experiment was

carried out with optimal values of the parameters (Run 12) However the predicted FLS of the

Run 11 has high relative error of 2278 Therefore the model shows itself with a low

predictive response for high Cr(III) feed concentrations

In order to surpass this problem improving thus the robustness and precision of the model

another regression equation incorporating the solute concentration as an additional independent

variable should be applied

Table 510 shows the predicted results by the regression model considering tads = (tE-tbp)

as the response variable It is visible that the relative errors of the predicted values are higher

than ones found by using the previous model As the inlet concentration increases the error

associated with the predicted value for tads also increase meaning that as the previous case

the feed concentration affects the response estimated by the model Relative error values of

2960 and 5434 were obtained for the Run 10 and Run 11 respectively

Table 510 Experimental values vs predicted values for the model built for the case in which the (tE-

tbp) was considered as dependent variable

Run CE

(mgL)

tads

experimental (min)

tads

predicted (min)

Relative error

()

10 202 16516 21404 2960

11 302 13868 21404 5434

12 100 19276 20784 1508

13 100 13272 14882 1213

55

For the case where it was used a feed concentration of 100 mgL Run 12 and Run 13 the

relative error between the predicted and experimental values is lower compared to the other

cases however it is still higher taking the value of 1508 and 1213 respectively

In short it can be concluded that both regression models appear to be very sensitive when

subjected to variations in operational parameters that were not taking into account in the

experimental design such as feed concentration However the model that considers FLS as

dependent variable provides better predicted values compared to the model in which tads was

the response variable

As previously mentioned greater concentrations cause a higher gradient leading to a faster

transport due to an increase diffusional coefficient Moreover given that fact the both models

were obtained from an experimental design in which the inlet concentration was not taken into

account as an independent variable it was expected to get predicted values with high associated

error since the feed concentration significantly influences the dynamics of the adsorption

process

57

6 CONCLUSION AND FUTURE WORK

This study demonstrated that the adsorption process operated in a fixed bed column packed

with mercerized pine bark have potential to be applied to the removal of Cr(III) ions from

wastewaters therefore be implemented in industrial processes

It was demonstrated that the application of a Box-Benken design is an efficient tool to assess

and optimize the process performance allowing a simultaneous evaluation of the response

variables (FLS and tads) when the bed height and superficial velocities were changed

The results obtained in the column studies showed that the uptake of chromium ions by

MPB was dependent on the bed depth superficial velocity and Cr(III) feed concentration It

was possible to observe that the adsorption capacity does not seem to vary significantly and the

maximum value obtained was 3535 mgg However the metal uptake increased with the

increasing of bed depth but decreased with the increase of the superficial velocity and inlet

concentrations Thus it can be concluded that for an efficient column utilization the adsorption

process should be operated at lower superficial velocities higher bed depths and lower feed

concentrations

A mathematical model that incorporates the LDF approximation for describing the

intraparticle transport was used to describe the dynamic behavior of the Cr(III) sorption in the

column tests A good agreement between model predictions and experimental data was

achieved

The statistical analysis of the results led to two regression models relating the independent

variables (bed height and superficial velocity) and the variable response (FLS or tads) It also

allowed to find which one of the operational parameters as more influence on the behavior of

the system according to the significance level obtained in the variance analysis The best results

were found with column operating at the following conditions H = 65 cm and uo =1443x10-2

cms whose responses were FLS = 65 and tads 133 min

The FLS predicted values were well predicted however for the case of the tads the

regression model was inaccurate since the predicted values were far from the experimental ones

The regeneration tests using two regenerative agents (NN-Bis and nitric acid) did not lead

to an efficient elution of Cr(III) ions from the adsorbent since the removal efficiencies achieved

were of 56 and 4685 with NN-Bis and HNO3 respectively

However since the adsorption process using fixed bed columns with agricultural residues

still in a phase of research and development technical and scientific issues should be solved to

bring this technology into commercialization

58

Based on this future perspectives can be made

bull It is still essential to prepare more efficiently the MPB

bull In order to obtain accurate results a real effluent should be used instead of a synthetic

one

bull Many mathematical models are used for a single metal adsorption process however the

elaboration of a new improved and simplified mathematical model is needed to get a

better response comparing to the experimental results

bull Further attention should be given to the application of adsorption technology using

fixed bed columns in product recovery-separation purification and recovery

bull At the industrial scale economic analyses are required to obtain the overall cost of the

adsorbent pre-treatment to purify a large quantity of wastewater

59

REFERENCES

1 Barnhart J (1997) Chromium chemistry and implications for environmental fate and

toxicity Journal of Soil Contamination 6 561-568

2 Fonseca-Correa R Giraldo L Moreno-Pirajaacuten JC (2013) Trivalent chromium

removal from aqueous solution with physically and chemically modified corncob waste

Journal of Analytical and Applied Pyrolysis 101 132-141

3 Kurniawan TA Chan GYS Lo WH Babel S (2006) Physico-chemical

treatment techniques for wastewater laden with heavy metals Chemical Engineering

Journal 118 83-98

4 Gupta VK Carrott PJM Ribeiro-Carrott MML Suhas (2009) Low-Cost

Adsorbents Growing Approach to Wastewater Treatmentmdasha Review Critical Reviews

in Environmental Science and Technology 39 783-842

5 Woodward RL (1970) Environmental pollution Canadian Medical Association

Journal 103 1316

6 Su C Jiang L Zhang W (2014) A review on heavy metal contamination in the soil

worldwide Situation impact and remediation techniques Environmental Skeptics

Critics 3 24-38

7 Gunatilake SK (2015) Methods of Removing Heavy Metals from Industrial

Wastewater Journal Multidisciplinary Engeineering Science Studies 1 2912-1309

8 Barakat MA (2011) New trends in removing heavy metals from industrial

wastewater Arabian Journal of Chemistry 4 361-377

9 Kotasacirc J Stasicka Z (2000) Chromium occurrence in the environment and methods

of its speciation Environmental Pollution 107 263ndash283

10 Ayangbenro AS Babalola OO (2017) A new strategy for heavy metal polluted

environments A review of microbial biosorbents International Journal of

Environmental Research and Public Health 14 94

11 Kumral E (2007) ldquoSpeciation of Chromium in Waters Via Sol-Gel Preconcentration

Prior To Atomic Spectrometric Determinationrdquo Master thesis submitted to School of

engineering and Sciences of Iacutezmir Institute of Technology

12 Vilar VJP Valle JAB Bhatnagar A et al (2012) Insights into trivalent

chromium biosorption onto protonated brown algae Pelvetia canaliculata Distribution

60

of chromium ionic species on the binding sites Chemical Engineering Journal 200-

202 140-148

13 Cossich ES Da Silva EA Tavares CRG Filho LC Ravagnani TMK (2004)

Biosorption of chromium(III) by biomass of seaweed Sargassum sp in a fixed-bed

column Adsorption 10 129-138

14 Fu F Wang Q (2011) Removal of heavy metal ions from wastewaters A review

Journal of Environmental Management 92 407-418

15 Lakherwal D (2014) Adsorption of Heavy Metals A Review International Journal

of Environmental Research and Development 4 41-48

16 OConnell DW Birkinshaw C amp ODwyer TF (2008) Heavy metal adsorbents

prepared from the modification of cellulose A review Bioresourcr Technology 99

6709-6724

17 Bruch LW Cole M W Zaremba E (1997) Physical Adsorption Forces and

Phenomena 1st Edition Clarendon Press (ISBN 0198556381)

18 Seader JD Ernest J and Henlet D K R (2013) Separation Process Principles 3rd

Edition John Wiley amp Sons (ISBN 9780470481837)

19 Ruthven DM (1984) Principles of Adsorption and Adsorption Processes Canada

John Wiley amp Sons (ISBN 0471866067)

20 Geankoplis CJ (1993) Transport processes and unit operations 3rd Edition Prentice-

Hall International Inc (ISBN 013045253)

21 Thomas WJ Crittenden B (1998) Adsorption Technology amp Design Elsevier

Science amp Technology Books (ISBN 0750619597)

22 Peacuterez-Mariacuten AB Aguilar MI Meseguer VF Ortuntildeo JF Saacuteez J Lloreacutens M

(2009) Biosorption of chromium (III) by orange (Citrus cinensis) waste Batch and

continuous studies Chemical Engineering Journal 155 199-206

23 Dabrowski A (2001) Adsorption - From theory to practice Advances in Colloid and

Interface Science 93 135-224

24 Ali I (2014) Water Treatment by Adsorption Columns Evaluation at Ground Level

Separation amp Purification Review 43 175-205

25 Ngah WS Hanafiah MAKM (2008) Removal of heavy metal ions from

wastewater by chemically modified plant wastes as adsorbents A review Bioresource

61

Technology 99 3935-3948

26 Ahmaruzzaman M (2011) Industrial wastes as low-cost potential adsorbents for the

treatment of wastewater laden with heavy metals Advances in Colloid and Interface

Science 166 36-59

27 Sud D Mahajan G Kaur MP (2008) Agricultural waste material as potential

adsorbent for sequestering heavy metal ions from aqueous solutions - A review

Bioresource Technology 99 6017-6027

28 Pettersen RC (1984) The chemical composition of wood The Chemistry of the solid

wood American Chemical Society Chapter 2 1-9 (ISBN 9780841207967)

29 Valentiacuten L Kluczek-Turpeinen B Willfoumlr S et al (2010) Scots pine (Pinus

sylvestris) bark composition and degradation by fungi Potential substrate for

bioremediation Bioresource Technology 101 2203-2209

30 Bagyaraj GRDJ (1982) Agricultural Microbiology New Delhi Prentice-Hall of

India (ISBN 9781483100333)

31 Thakur VK Thakur MK Gupta RK (2014) Review Raw Natural FiberndashBased

Polymer Composites International Journal of Polymer Analysis and Characterization

19 256-271

32 Wertz J L Beacutedueacute O and Mercier J P (2010) Cellulose Science and Technology

Lausanne EPFL Press (ISBN 978-1-4398-0799-6)

33 Oudiani A E Chaabouni Y Msahli S Sakli F (2011) Crystal transition from

cellulose i to cellulose II in NaOH treated Agave americana L fibre Carbohydrate

Polymers 86 1221-1229

34 Hill CAH (2006) Wood Modification Chemical Thermal and Other Processes

Christian V Stevens (ISBN 978-0-470-02172-9)

35 Miranda I Gominho J Mirra I Pereira H (2012) Chemical characterization of

barks from Picea abies and Pinus sylvestris after fractioning into different particle sizes

Industrial Crops Products 36 395-400

36 De Gisi S Lofrano G Grassi M Notarnicola M (2016) Characteristics and

adsorption capacities of low-cost sorbents for wastewater treatment A review

Sustainable Materials and Technology 9 10-40

37 Demirbas A (2008) Heavy metal adsorption onto agro-based waste materials A

62

review Journal of Hazardardous Materials 157 220-229

38 Arim AL Ceciacutelio DFM Quina MJ Gando-Ferreira LM (2018) Development

and characterization of pine bark with enhaced capacity for uptaking Cr(III) from

aqueous solutions Canadian Journal of Chemical Engineering (in press)

39 Ofomaja AE Naidoo EB Modise SJ (2009) Removal of copper(II) from aqueous

solution by pine and base modified pine cone powder as biosorbent Journal of

Hazardous Materials 168 909-917

40 Gurgel LVA Gil LF (2009) Adsorption of Cu(II) Cd(II) and Pb(II) from aqueous

single metal solutions by succinylated twice-mercerized sugarcane bagasse

functionalized with triethylenetetramine Water Research 43 4479-4488

41 McCabe W L Smith J Harriot P (1993) Unit operations of chemical engineering

5th Edition International Editions (ISBN0071127380)

42 Moreno-Pirajan J et al (2006) Scale-up of pilot plant for adsortion of heavy metals

An Asoc Quiacutem Argent 94 71-82

43 Green D W Perry R H (2007) Perrys Chemical Engineers Handbook 8th Edition

McGraw-Hill Education (ISBN 978-0071422949)

44 Ceciacutelio DFM (2016) Treatment of an effluent polluted with Cr ( III ) by adsorption

onto chemically modified agro-industrial waste Master thesis submitted to Department

of Chemical Engineering of University of Coimbra

45 Wankat P (1994) Rate-Controlled Separations Chapman and Hall (ISBN

9780751402841)

46 Xu Z Cai J Pan B (2013) Mathematically modeling fixed-bed adsorption in

aqueous systems Journal of Zhejiang University- Science A 14 155-176

47 Dym C (2004) Principles of Mathematical Modeling 2nd Edition Academic Press

(ISBN 9780080470283)

48 Du X Yuan Q Li Y (2008) Mathematical analysis of solanesol adsorption on

macroporous resins using the general rate model Chemical Engineering amp Technology

31 1310-1318

49 Eldestein-Keshet L (2005) Mathematical Models in Biology Society for Industrial and

Applied Mathematics Philadelphia (ISBN 0898715547)

50 De Almeida AM (2012) Estudo numeacuterico de colunas de adsorccedilatildeo de leito fixo

63

Master thesis submitted to University Federal do Paranaacute Sector of Technology

51 Hatzikioseyian A Mavituna F Tsezos M (1999) Modelling of fixed bed

biosorption columns in continuous metal ion removal processes The case of single solute

local equilibrium Process Metallurgy 9 429-448

52 Luz AD (2012) Aplicaccedilatildeo de coluna de adsorccedilatildeo em leito fixo para a remoccedilatildeo de

compostos BTX multicomponentes presentes em efluentes petroquiacutemicos Master

thesis submitted to Department of Chemical Engineering of University Federal de Santa

Catarina

53 Otero M Zabkova M Rodrigues AE (2005) Phenolic wastewaters purification by

thermal parametric pumping Modeling and pilot-scale experiments Water Resources

39 3467-3478

54 Glueckauf E (1955) Theory of chromatography Part 10 Formulae for diffusion into

spheres Trans Faraday Society 51 1540-1554

55 Gando-Ferreira LM Romatildeo IS Quina MJ (2011) Equilibrium and kinetic studies

on removal of Cu2+ and Cr3+ from aqueous solutions using a chelating resin Chemical

Engineering Journal 172 277-286

56 Chu KH (2004) Improved fixed bed models for metal biosorption Chememical

Engineering Journal 97 233-239

57 Tofan L Paduraru C Teodosiu C Toma O (2015) Fixed bed column study on

the removal of chromium ( III ) ions from aqueous solutions by using hemp fibers with

improved sorption performance Cellulose Chemistry and Technology 49 219-229

58 Calero M Hernaacuteinz F Blaacutezquez G Tenorio G Martiacuten-Lara MA (2009) Study

of Cr (III) biosorption in a fixed-bed column Journal of Hazardous Materials 171

886-893

59 Veit MT Silva EA Fagundes-Klen MR Tavares CRG Gonccedilalves GC (2008)

Biosorption of chromium(III) in fixed bed column Estudos tecnoloacutegicos 4 88-104

60 Tofan L Teodosiu C Paduraru C Wenkert R (2013) Cobalt (II) removal from

aqueous solutions by natural hemp fibers Batch and fixed-bed column studies Applied

Surface Science 285 33-39

61 Da Silva EA Cossich ES Tavares CRG Filho LC Guirardello R (2002)

Modeling of copper(II) biosorption by marine alga Sargassum sp in fixed-bed column

64

Process Biochemistry 38 791-799

62 Taty-Costodes VC Fauduet H Porte C Ho YS (2005) Removal of lead (II) ions

from synthetic and real effluents using immobilized Pinus sylvestris sawdust Adsorption

on a fixed-bed column Journal of Hazardous Materials 123 135-144

63 Malkoc E Nuhoglu Y(2005) Removal of Ni(II) ions from aqueous solutions using

waste of tea factory Adsorption on a fixed-bed column Journal of Hazardous

Materials 135 328-336

64 Tien C (1994) Adsorption Calculations and Modeling Boston Butterworth-

Heinemann (ISBN 0750691212)

65 Liu B Zeng L Mao J Ren Q (2010) Simulation of levulinic acid adsorption in

packed beds using parallel pore surface diffusion model Chemical Engineering amp

Technology 33 1146-1152

66 Ko DCK Porter JF McKay G (2005) Application of the concentration-dependent

surface diffusion model on the multicomponent fixed-bed adsorption systems

Chemical Engineering Science 60 5472-5479

67 Helfferich FG Carr PW (1993) Non-linear waves in chromatography I Waves

shocks and shapes Journal of Chromatogr 629 97-122

68 Chern J Chien Y (2001) Adsorption isotherms of benzoic acid onto activated carbon

and breakthrough curves in fixed-bed columns Water Research 36 3775-3780

69 Clark RM (1987) Evaluating the Cost and Performance of Field-Scale Granular

Activated Carbon Systems Environmental Science and Technology 1573-580

70 Hutchins R (1973) New method simplifies design of activated carbon systems

Chemical Engineering 80 133-138

71 YoonYH Nelson JH (1984) Application of Gas Adsorption Kinetics I A

Theoretical Model for Respirator Cartridge Service Life American Industrial Hygiene

Association Journal 45 509-16

72 Title S Title C et al (2013) Drug Delivery Systems Advanced Technologies

Potentially Applicable in Personalised Treatment 34

73 Brouwer P (2010) Theory of XRF 3rd Edition PANalytical BV The Netherlands

(ISBN 9090167587)

74 Heckert NA Filliben JJ Croarkin CM Hembree B Guthrie WF Tobias P

65

Prinz P (2002) Engineering Statistics e-Handbook

75 Han R Zou L Zhao X et al (2009) Characterization and properties of iron oxide-

coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column

Chemical Engineering Journal 149 123-131

I

7 APPENDIXES

71 Appendix A- Particle size distribution

In the Figure A1 it is possible to observe the particle size distribution of the MPB used in this

project

Figure A1 Experimental data for the particle size distribution of MPB

II

72 Appendix B- Experimental data

On the following Tables are present the experimental values obtained from the column

experiments for each case study

Table B1 Experimental data for the bed depth of 25 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 10 0 8 0

26 0 13 0 10 0

28 102 15 393 11 362

29 415 16 736 12 742

31 129 17 127 13 151

33 255 18 205 14 22

34 333 20 412 16 416

35 421 21 524 17 539

36 501 22 597 18 591

37 552 23 672 19 662

39 661 24 716 20 713

40 69 25 793 21 751

41 714 28 861 22 773

42 727 32 903 23 82

43 73 36 918 24 832

44 74 45 919 26 876

46 753 50 921 27 892

48 77 55 933 30 901

50 781 65 947 36 903

60 846 90 958 43 923

64 857 131 962 50 931

74 879 163 965 60 933

91 927 192 968 67 941

108 939 204 971 72 949

129 94 215 973 85 954

139 946 221 981 102 965

164 966 230 998 122 972

176 967 238 100 134 99

191 971 143 100

203 979

247 101

268 102

III

Table B2 Experimental data for the bed depth of 5 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

45 0 20 0 10 0

49 0 30 0 20 00505

50 133 31 182 21 316

51 323 32 45 22 653

52 662 33 728 23 115

53 116 34 113 24 178

55 196 35 183 25 301

56 269 36 282 26 428

57 333 37 349 27 551

58 394 38 445 28 627

59 472 39 531 29 677

60 566 40 598 30 699

61 611 41 634 31 739

63 659 42 658 33 771

65 698 43 699 34 797

67 711 47 793 35 833

72 757 52 839 36 86

74 795 60 864 39 875

79 812 71 881 42 887

89 818 76 894 50 905

103 838 81 90 53 907

143 887 90 915 56 912

185 923 100 923 69 913

207 924 120 935 83 928

227 931 133 948 92 941

246 958 168 962 120 947

270 971 179 974 134 953

295 100 201 101 143 958

305 996 217 100 169 97

226 101 181 981

206 998

217 100

224 101

IV

Table B3 Experimental data for the bed depth of 75 cm using different superficial velocities

962x10-2 cms 1443x10-2 cms 1924x10-2 cms

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

10 0 5 0 5 0

40 0 10 0 10 0

60 0 20 0 20 0

70 0 30 0 30 0

81 0778 38 0 32 0

84 604 43 0 34 122

86 117 49 211 36 677

88 182 52 378 38 146

90 248 54 51 40 267

92 357 56 659 43 55

94 453 93 844 44 635

96 542 98 85 46 712

98 601 116 852 48 738

100 66 134 852 50 751

104 697 165 863 52 762

110 717 182 88 54 77

140 81 198 891 56 781

147 824 209 926 87 862

159 866 223 967 113 876

180 892 231 975 123 889

202 893 139 925

225 90 160 945

237 914 175 971

257 923 196 983

300 937

315 938

336 939

345 945

360 958

V

Table B4 Experimental data for the bed depth of 75 cm and superficial velocity of 962x10-2 cms

using different feed concentrations

100 mgL 200 mgL 300 mgL

t (min) C (mgL) t (min) C (mgL) t (min) C (mgL)

0 0 0 0 0 0

70 0 38 0 24 0

81 0778 39 11 25 199

84 604 40 6 26 105

86 117 41 169 27 257

88 182 42 343 28 476

90 248 43 525 29 827

92 357 44 756 30 126

94 453 45 101 31 174

96 542 46 122 32 215

98 601 47 143 33 239

100 66 48 157 34 252

104 697 49 166 49 261

110 717 50 170 82 262

140 81 51 172 90 263

147 824 52 174 110 265

159 866 54 177 120 267

180 892 57 178 129 271

202 893 141 180 140 275

225 90 174 186 150 281

237 914 210 193 160 285

257 923 230 194 170 290

300 937 250 195 185 295

315 938 265 200 210 298

336 939 275 205 215 300

345 945 310 208 220 303

360 958 315 206 235 302

380 955

405 984

VI

Table B5 Experimental data for saturation of the column for the regeneration tests

962x10-2 cms 1443x10-2 cms

t (min) C (mgL) t (min) C (mgL)

0 0 0 0

50 0 52 0

60 0 54 257

62 214 56 866

63 559 59 234

64 834 62 463

65 134 64 606

67 28 66 666

68 332 74 747

69 414 100 823

70 491 125 867

72 706 146 91

73 763 177 93

74 784 195 969

120 82 209 973

151 823 235 101

165 834

190 89

205 897

231 929

248 946

285 972

302 981

317 104

335 103

355 102

VII

Table B6 Experimental data for the regeneration tests

N Nbis HNO3

t (min) C (mgL) t (min) C (mgL)

0 1080 0 100

35 366 05 908

4 268 1 536

45 21 15 619

5 1618 25 277

55 1532 3 214

6 1492 35 185

75 1178 4 170

9 1062 45 156

10 902 5 144

11 848 7 119

12 838 75 116

13 814 85 109

16 786 11 932

17 728 15 748

18 67 20 612

19 66 25 521

28 634 35 419

32 57 45 371

38 506 53 337

41 474 60 337

Page 15: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 16: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 17: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 18: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 19: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 20: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 21: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 22: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 23: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 24: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 25: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 26: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 27: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 28: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 29: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 30: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 31: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 32: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 33: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 34: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 35: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 36: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 37: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 38: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 39: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 40: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 41: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 42: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 43: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 44: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 45: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 46: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 47: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 48: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 49: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 50: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 51: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 52: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 53: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 54: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 55: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 56: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 57: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 58: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 59: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 60: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 61: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 62: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 63: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 64: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 65: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 66: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 67: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 68: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 69: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 70: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 71: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 72: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 73: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 74: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 75: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 76: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 77: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 78: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 79: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 80: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 81: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 82: Sorption of Cr(III) in fixed-bed column by modified pine ...
Page 83: Sorption of Cr(III) in fixed-bed column by modified pine ...